{
 "test": "ChatGPT prompts library, the-aischool.com/chatgpt-prompts",
 "date": "2026-10-28",
 "model": "claude-haiku-5-5 (Claude Haiku 5.5)",
 "how": "Claude Code 2.1.303 in print mode from an empty folder: claude -p <prompt> --model claude-haiku-5-5 --tools \"\" --strict-mcp-config --setting-sources \"\" --no-session-persistence --disable-slash-commands --effort low --output-format json < /dev/null. We did not sign in to ChatGPT. Every business, person and number in the prompts is made up.",
 "scoring": "format (yes, partial, no), length (yes, no, n.a.), invented facts (none, minor, major), gaps flagged where we left them on purpose (yes, no, n.a.); result: pass, minor (one small slip), fail (breaks a rule the prompt states). Scored by hand.",
 "cost_note": "cost_usd is the CLI's own total_cost_usd at list price, which includes Claude Code's own instructions sent with every request.",
 "summary": {
  "runs": 50,
  "prompts": 24,
  "jobs": 11,
  "round1": {
   "pass": 13,
   "minor": 6,
   "fail": 5
  },
  "final_versions": {
   "pass": 23,
   "minor": 9,
   "fail": 1
  },
  "changed_prompts": [
   "T03",
   "T05",
   "T07",
   "T10",
   "T18",
   "T20",
   "T22"
  ],
  "cost_usd_total": 0.04002,
  "cost_usd_per_run_min": 0.00027041,
  "cost_usd_per_run_max": 0.0019400799999999998,
  "seconds_per_run_min": 2.216,
  "seconds_per_run_max": 19.441,
  "seconds_per_run_median": 9.956,
  "translated_runs": 6,
  "translated": {
   "pass": 3,
   "minor": 3,
   "fail": 0
  },
  "translated_cost_usd": 0.00468,
  "models": [
   "claude-haiku-5-5"
  ]
 },
 "runs": [
  {
   "case": "T01",
   "locale": "en",
   "prompt_id": "reply-difficult-email",
   "job": "writing-email",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a reply to the email below. I am the owner of Lakeside Bakery, a small bakery that makes custom cakes and the sender is Priya, a customer who ordered a birthday cake.\nWhat I want to happen next: she accepts a replacement cake or a refund and feels heard.\nWhat I can offer: a free replacement cake delivered by 2 pm this Saturday, or a full refund of $64.\nTone: calm, direct and polite. Under 150 words. Do not promise anything beyond what I can offer.\nIf the email asks something I have not answered above, write [CHECK: the question] in that spot instead of guessing.\n\nEmail:\nHi, I picked up my daughter's birthday cake yesterday and the name was spelled wrong (Mia instead of Maya), and half the frosting had slid off by the time we got home. The party was a mess. I want to know how this happened, and whether you will also cover the $30 I spent on supermarket cupcakes as a backup. Also, do you deliver to Brookfield?\nPriya",
   "answer": "Subject: Re: Your birthday cake order\n\nHi Priya,\n\nI'm sorry the cake was wrong and that it arrived damaged. Your daughter's birthday should not have gone like that, and I understand how stressful the party was.\n\nI'd like to make this right in one of two ways:\n\n- **A free replacement cake**, with the name spelled correctly, delivered by 2 pm this Saturday.\n- **A full refund of $64.**\n\nPlease reply with the option you prefer and I'll arrange it right away.\n\n[CHECK: how the spelling error happened]\n\n[CHECK: how the frosting came to slide off]\n\n[CHECK: whether you will cover the $30 you spent on supermarket cupcakes]\n\n[CHECK: whether you deliver to Brookfield]\n\nSincerely,\n[Your name]\nLakeside Bakery",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "117 words of 150. Offered only the replacement or the $64 refund. Four [CHECK] lines for the unanswered questions (how it happened, the $30 cupcakes, Brookfield delivery), placed at the end rather than in the text.",
   "seconds": 7.8,
   "cost_usd": 0.000485
  },
  {
   "case": "T02",
   "locale": "en",
   "prompt_id": "email-from-notes",
   "job": "writing-email",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Turn my notes below into an email to Daniel, our bookkeeper about the overdue October invoices.\nThe goal of the email: get him to send three overdue invoice reminders this week.\nUnder 120 words, in the order that makes the request clearest, ending with one clear next step.\nUse only what is in my notes. If a note is unclear or a detail is missing, write [CHECK] there and keep going.\n\nNotes:\n- 3 clients overdue: Harbor Dental (invoice 1043, $1,200), Pine & Co (1047), Westgate Gym (1051, $860)\n- Harbor Dental is 45 days late, the other two about 20\n- use the friendly reminder template, not the final notice one\n- cc me on Harbor Dental\n- new late-fee policy?? not sure if it starts in Nov or Jan",
   "answer": "**Subject:** Reminders needed for three overdue October invoices\n\nHi Daniel,\n\nCould you send friendly payment reminders for these three overdue invoices this week?\n\n- Harbor Dental, invoice 1043, $1,200 (45 days late; please cc me)\n- Pine & Co, invoice 1047 (about 20 days late)\n- Westgate Gym, invoice 1051, $860 (about 20 days late)\n\nPlease use the friendly reminder template, not the final notice.\n\nOn the new late-fee policy, [CHECK: does it start in November or January?]. Should I mention it in these reminders?\n\nCould you confirm once all three are sent?\n\nThanks,\n[CHECK: your name]",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "93 words of 120. Left the Pine & Co amount out instead of inventing it and put [CHECK] on the late-fee start date.",
   "seconds": 7.6,
   "cost_usd": 0.000646
  },
  {
   "case": "T03",
   "locale": "en",
   "prompt_id": "product-description",
   "job": "marketing-social",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a product description for the Fjord Wool Throw blanket for my Shopify product page.\nFacts: 100% merino wool, 130 x 170 cm, $89, machine washable on the wool cycle at 30 C, made in Portugal, comes rolled in a cotton bag, colors oat, slate and moss\nIdeal buyer: people furnishing a small apartment who want one warm, good-looking throw for the sofa\nFormat: a benefit headline of one line, a paragraph of 50 to 80 words, then 4 bullets that each pair a feature with what it means for the buyer.\nDo not invent certifications, awards, reviews, origins or measurements.",
   "answer": "**Cozy Up in Pure Merino: The Fjord Wool Throw**\n\nMake your small-space sofa feel like the coziest spot in the home. The Fjord Wool Throw is woven from 100% merino wool, so it's soft against the skin and warm without weighing you down. At 130 x 170 cm, it's generous enough to wrap around you on movie nights yet compact enough to fold neatly over an armrest. Made in Portugal and available in oat, slate, and moss, it brings warmth and quiet style to any living room. It machine washes on the wool cycle at 30°C, so keeping it fresh is simple. It arrives rolled in a cotton bag, ready to gift or store. Priced at $89.\n\n- **100% merino wool:** Soft, warm, and gentle on skin, so it feels comfortable from the first cold evening onward.\n- **130 x 170 cm size:** Covers your lap and shoulders on the sofa without overwhelming a compact space.\n- **Machine washable on the wool cycle at 30°C:** Freshening it up takes minutes at home, with no dry cleaning trips.\n- **Oat, slate, and moss colors:** Easy to match with your existing décor, from light neutrals to deeper, earthy palettes.",
   "format": "yes",
   "length": "no",
   "invented": "none",
   "missing": "n.a.",
   "result": "fail",
   "note": "The paragraph ran 108 words against a limit of 80. No invented certifications or measurements.",
   "seconds": 5.8,
   "cost_usd": 0.00046
  },
  {
   "case": "T04",
   "locale": "en",
   "prompt_id": "ad-variations",
   "job": "marketing-social",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write 4 versions of a Meta ad for a 6-week beginner pottery course on Tuesday evenings, $240 with materials included aimed at adults in Portland who want a screen-free hobby.\nEach version: primary text under 125 characters, a headline under 40 characters, and its angle in brackets: pain, result, proof or objection.\nUse only these facts and proof: classes of 8 people at most; 4.9 stars from 112 Google reviews; you take home 4 to 6 finished pieces; the next course starts November 18.\nNo \"best\", \"number one\" or other superlatives unless my facts prove them.",
   "answer": "**Version 1 [Pain]**\n- **Primary text:** Scrolling again tonight? Trade your phone for clay. 6 Tuesday evenings, classes of 8 max, in Portland.\n- **Headline:** A screen-free hobby in Portland\n\n**Version 2 [Result]**\n- **Primary text:** Six Tuesday evenings and you'll leave with 4 to 6 finished pieces you made yourself. Materials included.\n- **Headline:** Take home 4 to 6 finished pieces\n\n**Version 3 [Proof]**\n- **Primary text:** Rated 4.9 stars from 112 Google reviews. Beginner pottery, Tuesday evenings, classes of 8 at most.\n- **Headline:** 4.9 stars from 112 Google reviews\n\n**Version 4 [Objection]**\n- **Primary text:** Worried you're not creative? You don't need experience. $240 covers the course and materials. Next starts November 18.\n- **Headline:** $240 with materials included",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Primary texts 98 to 118 characters (limit 125), headlines 28 to 33 (limit 40), angle labelled on all 4. 'You don't need experience' is read from 'beginner course'.",
   "seconds": 9.7,
   "cost_usd": 0.000909
  },
  {
   "case": "T05",
   "locale": "en",
   "prompt_id": "cold-email",
   "job": "sales",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a cold email to a office manager at a dental practice with 2 to 5 dentists.\nWhat I sell: a text-message service that confirms appointments 48 hours ahead. The problem it solves for them: no-shows and last-minute cancellations that leave chairs empty.\nProof I can mention: none yet. The ask: a 15-minute call next week.\nUnder 120 words, plain text, no flattering opener and no fake familiarity.\nLeave [PERSONAL LINE] where I will add one real detail about them. Give two subject lines.",
   "answer": "**Subject line options:**\n\n1. Empty chairs from cancellations?\n2. Cutting no-shows at [PERSONAL LINE]\n\n---\n\nHi [First Name],\n\nMost dental offices lose a few hours a week to patients who don't show up or cancel last minute, and those slots rarely get refilled.\n\nWe run a text-message service that confirms each appointment 48 hours ahead. Patients reply to confirm or reschedule, so your team hears about cancellations early enough to fill the chair.\n\n[PERSONAL LINE]\n\nWould you be open to a 15-minute call next week? I can show you how it works and what it would look like for your practice.\n\nBest,\n[Your Name]",
   "format": "yes",
   "length": "yes",
   "invented": "major",
   "missing": "n.a.",
   "result": "fail",
   "note": "103 words of 120, but with proof 'none yet' it invented a feature (patients reply to confirm or reschedule) and an industry claim (offices 'lose a few hours a week').",
   "seconds": 4.9,
   "cost_usd": 0.000356
  },
  {
   "case": "T06",
   "locale": "en",
   "prompt_id": "proposal-summary",
   "job": "sales",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a one-page proposal summary for Greenline Landscaping for a new 6-page website.\nTheir goal, in their words: \"We want more quote requests from homeowners, not just calls asking for prices.\"\nIncluded: 6 pages (home, services, about, gallery, reviews, contact), a quote request form that emails the office, mobile-friendly design, basic on-page SEO, 2 rounds of revisions\nNot included: writing copy for more than 6 pages, photography, paid ads, hosting fees\nTimeline: 5 weeks from the kickoff call, which is on November 10, 2026\nPrice and terms: $4,800, 50% at kickoff and 50% at launch\nNext step: sign the proposal and book the kickoff call by November 5\nUse these headings: Your goal, What we will do, Not included, Timeline, Investment, Next step.\nUnder 350 words. No guarantees of results.",
   "answer": "# Website Proposal for Greenline Landscaping\n\n**Your goal**\nYou want more quote requests from homeowners, not just calls asking for prices. This project is built to turn your website into a place where homeowners can see your work, trust your team, and request a quote in a few minutes.\n\n**What we will do**\n- Build a 6-page website: Home, Services, About, Gallery, Reviews, and Contact.\n- Create a quote request form that emails your office directly, so every request reaches your team.\n- Design for mobile first, since most homeowners will find you on their phones.\n- Apply basic on-page SEO, including page titles, descriptions, headings, and local service keywords, to help local homeowners find your pages.\n- Include 2 rounds of revisions to refine the design and content.\n\n**Not included**\n- Copywriting for more than the 6 pages listed above\n- Photography\n- Paid advertising\n- Hosting fees\n\n**Timeline**\nFive weeks from the kickoff call on November 10, 2026, which puts launch around December 15, 2026.\n\n**Investment**\n$4,800 total\n- 50% ($2,400) due at kickoff\n- 50% ($2,400) due at launch\n\n**Next step**\nSign this proposal and book the kickoff call by November 5, 2026.",
   "format": "yes",
   "length": "yes",
   "invented": "minor",
   "missing": "n.a.",
   "result": "minor",
   "note": "184 words of 350, the six headings in order, prices and dates exact, launch date worked out correctly as December 15. Added one unsupported line: most homeowners will find you on their phones.",
   "seconds": 6.5,
   "cost_usd": 0.000573
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe. Offer nothing beyond the remedy above.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Dear [Name],\n\nThree outages this month, and a lost day of client calls on Monday, is a serious failure on our side. You also should not have had to chase us after your last ticket. I'm sorry.\n\nYour next bill will include a credit for three days of service. A technician will visit this Thursday between 8 am and noon at no charge. They'll test your line and equipment for the fault behind these drops and fix it during the visit if it's on our network. Please reply to confirm someone will be home during that window.\n\n[Your name]\nBrightwave Internet\n\n---\n\nI didn't address the $400 request or the switching threat, since the remedy you gave me doesn't cover either. If you want to acknowledge them, you'll need to decide that separately.",
   "format": "yes",
   "length": "yes",
   "invented": "minor",
   "missing": "n.a.",
   "result": "minor",
   "note": "About 100 words for the reply (132 with its note to us). Did not offer the $400, but promised the technician would fix the fault during the visit if it was on the network.",
   "seconds": 10.3,
   "cost_usd": 0.000831
  },
  {
   "case": "T08",
   "locale": "en",
   "prompt_id": "faq-from-tickets",
   "job": "customer-support",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Here are recent customer questions: 1. Can I try a class before buying a pass?\n2. How much is a monthly membership?\n3. Do you have mats or do I bring my own?\n4. Is there parking near the studio?\n5. Can I freeze my membership when I travel?\n6. What is the difference between flow and yin?\n7. Do I need to book ahead?\n8. Is the first class free?\n9. What happens if I cancel late?\n10. Do you offer prenatal yoga?\n11. Can I share my pass with my partner?\n12. How early should I arrive?.\nGroup them into the most common topics, up to 8. For each topic, write the question the way customers phrase it and an answer of 2 to 4 sentences using only these facts: Studio: Willow Yoga, 14 Elm Street. Drop-in class $22. 10-class pass $180, valid 4 months, for one person only. Unlimited monthly membership $129, can be frozen once a year for up to 30 days. First class for new students is $10. Mats are free to borrow. Booking through the app is required; classes hold 18 people. Cancel at least 6 hours before class or the class is charged. Flow classes are active and move with the breath; yin classes hold poses for 3 to 5 minutes. Arrive 10 minutes early for your first class..\nIf my facts do not answer a question, write [NEEDS ANSWER] instead of making one up.",
   "answer": "**1. Trying a class before committing**\n*\"Can I try a class before buying a pass?\" / \"Is the first class free?\" / \"How early should I arrive?\"*\n\nYes, new students can take a first class for $10. It isn't free, but it's a low-cost way to try us before buying a pass. If you want to keep coming without a pass, drop-in classes are $22 each. Please arrive 10 minutes early for your first class.\n\n**2. Membership and pass prices**\n*\"How much is a monthly membership?\" / \"Can I share my pass with my partner?\"*\n\nThe unlimited monthly membership is $129. The 10-class pass is $180 and is valid for 4 months. Passes are for one person only, so your partner will need their own.\n\n**3. Mats**\n*\"Do you have mats, or do I bring my own?\"*\n\nYes, we have mats. You don't need to bring your own, since mats are free to borrow.\n\n**4. Parking**\n*\"Is there parking near the studio?\"*\n\n[NEEDS ANSWER]\n\n**5. Freezing a membership**\n*\"Can I freeze my membership when I travel?\"*\n\nYes, you can freeze your unlimited monthly membership once a year for up to 30 days. The freeze can only be used once per year, so plan any trips accordingly.\n\n**6. Flow vs. yin**\n*\"What is the difference between flow and yin?\"*\n\nFlow classes are active and move with your breath. Yin classes hold each pose for 3 to 5 minutes, making them a slower, more still practice. Choose flow for movement and energy, or yin for a calmer session.\n\n**7. Booking and cancellations**\n*\"Do I need to book ahead?\" / \"What happens if I cancel late?\"*\n\nYes, booking through the app is required, and classes hold 18 people. If you cancel at least 6 hours before class, you won't be charged. If you cancel later than that, the class will be charged.\n\n**8. Prenatal yoga**\n*\"Do you offer prenatal yoga?\"*\n\n[NEEDS ANSWER]\n\nTwo questions need answers from you: parking and prenatal yoga.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "8 topics, answers of 2 to 4 sentences from the facts only, [NEEDS ANSWER] for parking and prenatal yoga, and 'the first class is not free, it is $10'.",
   "seconds": 17.3,
   "cost_usd": 0.00194
  },
  {
   "case": "T09",
   "locale": "en",
   "prompt_id": "title-meta",
   "job": "seo-content",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write 3 title tags of 50 to 60 characters and 3 meta descriptions of 140 to 155 characters for the page below.\nMain keyword: bathroom remodel cost. What the page offers that others do not: real price ranges from 38 bathroom remodels we finished in Austin in 2025 and 2026.\nShow the character count after each one.\n\nPage:\nOur guide to bathroom remodel cost in Austin. Across 38 jobs we completed in 2025 and 2026, a cosmetic refresh (paint, fixtures, vanity) cost $9,500 to $16,000, a mid-range remodel with new tile and a walk-in shower cost $22,000 to $38,000, and a full gut with layout changes cost $45,000 to $70,000. The page explains what drives the price, how long each type takes and which permits Austin requires.",
   "answer": "**Title tags**\n\n1. Bathroom Remodel Cost in Austin: Real 2025-2026 Prices (54 characters)\n2. Bathroom Remodel Cost Austin: 38 Real Projects Priced (53 characters)\n3. Austin Bathroom Remodel Cost: Actual 2025-2026 Ranges (53 characters)\n\n**Meta descriptions**\n\n1. Bathroom remodel cost in Austin from 38 jobs we finished in 2025 and 2026. Price ranges for cosmetic, mid-range and full gut projects, plus permits. (150 characters)\n2. We finished 38 bathroom remodels in Austin in 2025 and 2026. Real costs for cosmetic, mid-range and full gut jobs, with timelines and permit requirements. (153 characters)\n3. Austin bathroom remodel prices from 38 finished jobs. Cosmetic, mid-range and full gut ranges, what drives cost, and which permits Austin requires. (147 characters)\n\nEach count includes spaces and punctuation, and the en dash in the 2025-2026 titles counts as one character.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "minor",
   "note": "Titles 53 to 54 characters, descriptions 147 to 154: all inside the limits. Two of its six counts were off by 1 or 2 characters (said 150 for 148, 153 for 154). Used an en dash in '2025-2026'.",
   "seconds": 15.3,
   "cost_usd": 0.001747
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nGive: a title under 60 characters with the keyword near the start, a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].",
   "answer": "# How to Price Handmade Jewelry: A Formula That Covers Materials, Time, and Fees\n\n**Title (under 60 characters):** How to Price Handmade Jewelry (Without Underselling)\n*(63 characters; shortened option: \"How to Price Handmade Jewelry Profitably\" - 43 characters)*\n\n**Direct answer (opening 2 sentences):**\nTo price handmade jewelry, add the cost of materials, your hourly labor rate multiplied by the time each piece takes, and the platform and payment fees, then apply a markup that covers overhead and profit. Most makers land on a final price by multiplying that base cost by 2 to 3, then checking it against what similar pieces sell for on Etsy and at local craft fairs.\n\n---\n\n## H2: What Materials Should I Include in My Jewelry Cost?\n\n- List every component: wire, beads, clasps, jump rings, chain, findings, earring hooks, pads, and packaging\n- Explain how to calculate cost per unit (bulk price ÷ quantity) rather than what you paid for the whole pack\n- Include waste and breakage (add 5-10% to material costs)\n- Show a worked example for one simple piece\n- **[MY PROOF]:** Photo of your actual material receipts or a supplier invoice showing bulk pricing\n- **[MY PROOF]:** A breakdown of one of your best-selling pieces, itemized by component with real costs\n\n## H2: How Do I Calculate My Labor Cost Per Piece?\n\n- Explain tracking time from start to finish, including design, cutting, assembly, finishing, and photos\n- Set an hourly rate (suggest starting at a set minimum you can live on, then adjusting)\n- Show the formula: minutes per piece ÷ 60 × hourly rate\n- Note that first-time designs take longer and should be timed separately from repeat production\n- **[MY PROOF]:** Screenshot of a time-tracking app or spreadsheet logging real production hours\n- **[MY PROOF]:** Your own stated hourly rate and how you arrived at it\n\n## H2: What Overhead Costs Should I Add to Each Piece?\n\n- Cover tools, equipment wear, workspace, electricity, and software\n- Explain how to spread monthly overhead across expected monthly sales\n- Give a simple example: $120 monthly overhead ÷ 60 pieces sold = $2 per piece\n- **[MY PROOF]:** Your actual monthly overhead list (tools, subscriptions, booth fees) with real numbers\n\n## H2: How Much Do Etsy and Craft Fair Fees Add to My Price?\n\n- Break down Etsy listing fees, transaction fees, payment processing, and offsite ad fees\n- Explain craft fair booth fees, table costs, and card reader fees as per-piece costs\n- Show how to build fees into the price so they don't come out of your profit\n- Provide a formula: (Base cost + fees) ÷ (1 − fee percentage)\n- **[MY PROOF]:** Screenshot of your Etsy fee statement from the last 30 days\n- **[MY PROOF]:** Your craft fair booth invoice and sales count from one event\n\n## H2: What Markup Should I Apply to Handmade Jewelry?\n\n- Explain why a 2x-3x markup on total cost is a common starting point for handmade goods\n- Discuss how retail pricing differs from wholesale (wholesale is typically 50% of retail)\n- Note that markup must cover profit, not just break even\n- Warn against pricing purely on materials cost\n- **[MY PROOF]:** Your own markup calculation for a piece you currently sell, with before-and-after numbers\n\n## H2: How Do I Check If My Price Will Actually Sell?\n\n- Explain how to research competitor prices on Etsy using filters for handmade and similar materials\n- Compare against craft fair price points in your region\n- Discuss price perception: buyers judge quality by price signals and presentation\n- Advise testing a price increase on one piece before changing the whole line\n- **[MY PROOF]:** Screenshots of 3-5 comparable listings from Etsy, with your notes on how your piece differs\n- **[MY PROOF]:** Sales data showing how a past price change affected units sold\n\n## H2: How Do I Put the Full Pricing Formula Together?\n\n- Walk through the complete formula step by step: materials + labor + overhead + fees = base cost, then apply markup\n- Use one worked example from start to finish, with a table of each cost line\n- Provide a printable or copyable pricing worksheet template\n- Explain how to review and update prices every 3-6 months as material costs change\n- **[MY PROOF]:** A completed worksheet using one of your own pieces, with real numbers filled in\n- **[MY PROOF]:** A photo of the finished piece with its final listing price\n\n## H2: What Mistakes Should I Avoid When Pricing Jewelry?\n\n- Forgetting to count your own time\n- Ignoring platform and payment fees\n- Discounting too deeply during sales, which erodes margin\n- Pricing based on what you feel comfortable charging rather than the formula\n- Not accounting for shipping, packaging, or free-shipping offers\n- **[MY PROOF]:** An example of a pricing mistake you made (or saw) and what it cost in real terms\n\n---\n\n**Suggested closing:** Recap the formula in one line, and invite readers to download the worksheet or share their own pricing questions.",
   "format": "partial",
   "length": "no",
   "invented": "major",
   "missing": "n.a.",
   "result": "fail",
   "note": "Opened with a 78-character headline, then called a 52-character title '63 characters'. Stated rules of thumb as facts: a 2x to 3x markup, wholesale at 50% of retail, 5 to 10% waste.",
   "seconds": 13.4,
   "cost_usd": 0.001147
  },
  {
   "case": "T11",
   "locale": "en",
   "prompt_id": "summarize-for-decision",
   "job": "research-summaries",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Summarize the document below for our operations director, who needs to decide whether to renew our office cleaning contract with CleanCo for 2027.\nGive: a 3-sentence summary, the 5 points that matter for the decision, numbers and dates exactly as written, and the questions the document leaves open.\nSay which section each point comes from. Add nothing that is not in the document.\n\nDocument:\nCleanCo service review, January to September 2026\n\n1. Scope. CleanCo cleans our two floors (about 1,100 square meters) five evenings a week. The contract costs $3,450 a month and ends on December 31, 2026. Renewal requires 60 days' notice, so the deadline to decide is November 1, 2026.\n\n2. Quality. Facilities logged 41 complaints in nine months: 23 about kitchen and bathroom cleaning, 12 about missed trash collection and 6 about noise after 9 pm. Complaints fell from 9 in March to 2 in September after CleanCo replaced its night supervisor in May.\n\n3. Incidents. On July 8 a cleaner left the server room door propped open overnight. No equipment was harmed. CleanCo retrained the crew and added a door checklist.\n\n4. Price for 2027. CleanCo has quoted $3,690 a month for 2027, a 7% increase, citing wage rises. They offer to keep the 2026 price if we sign for two years.\n\n5. Alternatives. Facilities requested one competing quote, from ShineWorks: $3,300 a month for the same scope, with a 90-day trial clause. ShineWorks has not visited the site yet and has not provided references.",
   "answer": "**Summary**\n\nCleanCo has cleaned the two floors (about 1,100 square meters) five evenings a week in 2026 at $3,450 a month, and its contract ends December 31, 2026. For 2027, CleanCo has quoted $3,690 a month, and the only competing quote, from ShineWorks, is $3,300 a month. Renewal requires 60 days' notice, so the decision is due by November 1, 2026.\n\n**Five points for the decision**\n\n1. **Renewal deadline (Section 1, Scope):** Renewal requires 60 days' notice, so the deadline to decide is November 1, 2026. The contract ends December 31, 2026.\n2. **CleanCo's 2027 price (Section 4, Price for 2027):** CleanCo quoted $3,690 a month, a 7% increase, citing wage rises. It offers to keep the 2026 price if we sign for two years.\n3. **Service quality (Section 2, Quality):** Facilities logged 41 complaints in nine months: 23 about kitchen and bathroom cleaning, 12 about missed trash collection and 6 about noise after 9 pm. Complaints fell from 9 in March to 2 in September, after CleanCo replaced its night supervisor in May.\n4. **Security incident (Section 3, Incidents):** On July 8, a cleaner left the server room door propped open overnight. No equipment was harmed. CleanCo retrained the crew and added a door checklist.\n5. **Alternative quote (Section 5, Alternatives):** ShineWorks quoted $3,300 a month for the same scope, with a 90-day trial clause. However, ShineWorks has not visited the site yet and has not provided references.\n\n**Open questions the document does not answer**\n\n- Does the 2026 price hold only for a two-year signing, and what are the terms of that offer?\n- Will complaints stay low after September? The document's data ends in September 2026.\n- Does ShineWorks's $3,300 quote cover the same service quality as CleanCo's, and what are the terms of its 90-day trial clause?\n- Can ShineWorks provide references, and will it visit the site before the November 1 deadline?\n- Is the 7% increase tied to specific wage data, and what does CleanCo's one-year (non-two-year) price look like?",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "3-sentence summary, 5 points each with its section, every number and date exact, 5 open questions, nothing added.",
   "seconds": 7.5,
   "cost_usd": 0.000986
  },
  {
   "case": "T12",
   "locale": "en",
   "prompt_id": "compare-options",
   "job": "research-summaries",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Compare PrintHub, InkWorks and Papercraft Co for printing 500 tri-fold brochures for a trade show on December 3.\nMy criteria, most important first: delivery before December 1, total price, paper quality, a printed proof before the full run.\nUse only the information I paste below, and write \"not stated\" where it is missing.\nReturn a table, then a recommendation of up to 3 sentences on which option fits which situation.\n\nInformation:\nPrintHub quote: 500 tri-fold brochures, 150 gsm gloss, $412 including shipping, delivery 6 business days after proof approval, PDF proof by email within 24 hours.\nInkWorks quote: 500 brochures, 170 gsm silk, $389 plus $35 shipping, delivery November 26 if we order by November 14. No mention of proofs.\nPapercraft Co email: 500 brochures on 200 gsm recycled matte, $520 total, printed proof mailed to us before the full run. Delivery date to be confirmed after artwork review.",
   "answer": "| Criterion (in your priority order) | PrintHub | InkWorks | Papercraft Co. |\n|---|---|---|---|\n| 1. Delivery before Dec 1 | Not stated as a date. Delivery is 6 business days after proof approval. | Nov 26, if ordered by Nov 14. | Not stated. To be confirmed after artwork review. |\n| 2. Total price | $412, including shipping. | $424 ($389 + $35 shipping). | $520 total. |\n| 3. Paper quality | 150 gsm gloss. | 170 gsm silk. | 200 gsm recycled matte. |\n| 4. Printed proof before full run | No. PDF proof by email within 24 hours. | Not stated. No mention of proofs. | Yes. Printed proof mailed before the full run. |\n\n**Recommendation:** InkWorks fits if you can place the order by November 14, since it is the only option with a stated delivery date before December 1, but its proof terms are not stated. PrintHub fits if you can approve a PDF proof quickly and accept that delivery depends on approval timing, since it is the lowest total price. Papercraft Co. fits only if a printed proof matters more to you than price and a confirmed delivery date.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "Table plus a 3-sentence recommendation. 'Not stated' for the InkWorks proof and the Papercraft delivery date; added InkWorks shipping correctly to $424.",
   "seconds": 10.4,
   "cost_usd": 0.001039
  },
  {
   "case": "T13",
   "locale": "en",
   "prompt_id": "sheet-formula",
   "job": "spreadsheets-data",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a formula for Google Sheets that adds up the order totals only for orders that are marked Paid and were placed in October 2026.\nMy data: column A has the order date, column D has the order total in dollars, column E has the status (Paid, Pending or Refunded), rows 2 to 400, headers in row 1.\nGive the formula, explain each part in one line, and show the result it gives for these sample rows: A2 = 2026-10-14, D2 = 129.50, E2 = Paid; A3 = 2026-10-20, D3 = 80.00, E3 = Pending; A4 = 2026-09-30, D4 = 45.00, E4 = Paid; A5 = 2026-10-31, D5 = 60.25, E5 = Paid.",
   "answer": "**Formula:**\n\n```\n=SUMIFS(D2:D400, E2:E400, \"Paid\", A2:A400, \">=\"&DATE(2026,10,1), A2:A400, \"<\"&DATE(2026,11,1))\n```\n\n**Explanation:**\n\n- `SUMIFS(` adds up values in column D that meet every condition listed after it.\n- `D2:D400` is the order totals, the values to add.\n- `E2:E400, \"Paid\"` keeps only rows where the status is exactly \"Paid\".\n- `A2:A400, \">=\"&DATE(2026,10,1)` keeps rows dated on or after October 1, 2026.\n- `A2:A400, \"<\"&DATE(2026,11,1)` keeps rows dated before November 1, 2026, so October 31 is included.\n\n**Result for the sample rows:** **189.75**\n\n- Row 2 (Oct 14, Paid, 129.50): included\n- Row 3 (Oct 20, Pending): excluded, not Paid\n- Row 4 (Sep 30, Paid): excluded, wrong month\n- Row 5 (Oct 31, Paid, 60.25): included\n\n129.50 + 60.25 = 189.75\n\nThis assumes column A holds real dates. If the dates were pasted as text, the date comparisons will not match and the result will be 0. In that case, tell me and I'll adjust the formula.",
   "format": "yes",
   "length": "n.a.",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Correct SUMIFS with date bounds; sample result 189.75 is right; one line per part.",
   "seconds": 6.3,
   "cost_usd": 0.000591
  },
  {
   "case": "T14",
   "locale": "en",
   "prompt_id": "clean-list",
   "job": "spreadsheets-data",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Clean this list: name | phone | signup date\n  maria LOPEZ | (555) 010-4417 | 3/7/2026\nJAMES o'neil | 555.010.2290 | 2026-03-09\n  li wei  | | March 12 2026\nSofia  Rossi | 5550109981 | 12/03/2026\ntom baker | 555-010-77 | 3/15/26\nAisha Khan | +1 555 010 6623 | 2026/03/18.\nRules: trim spaces, names in Title Case, US phone numbers as +1 555 010 0100, dates as YYYY-MM-DD.\nReturn it as a table I can paste into a spreadsheet, then list every row you could not clean and why.\nDo not guess missing values; leave them empty.",
   "answer": "| name | phone | signup date |\n|---|---|---|\n| Maria Lopez | +1 555 010 4417 | 2026-03-07 |\n| James O'Neil | +1 555 010 2290 | 2026-03-09 |\n| Li Wei | | 2026-03-12 |\n| Sofia Rossi | +1 555 010 9981 | |\n| Tom Baker | | 2026-03-15 |\n| Aisha Khan | +1 555 010 6623 | 2026-03-18 |\n\n**Rows I could not fully clean:**\n\n- **Li Wei:** No phone number was given, so the phone cell is empty.\n- **Sofia Rossi:** The date `12/03/2026` could be December 3 or March 12, and the list doesn't settle it. The date cell is empty. Li Wei's row says \"March 12 2026,\" which suggests March 12, but I didn't use that to fill it in.\n- **Tom Baker:** `555-010-77` has only 7 digits, so it isn't a complete US number. The phone cell is empty.\n\nI read slash dates as month/day/year, since `3/15/26` only works in that order. If you want `3/7/2026` read as day/month instead, tell me and I'll fix that row.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "6 rows in, 6 out. Left Li Wei's phone, Tom Baker's short number and Sofia Rossi's ambiguous 12/03/2026 empty and explained each.",
   "seconds": 12.3,
   "cost_usd": 0.001019
  },
  {
   "case": "T15",
   "locale": "en",
   "prompt_id": "numbers-story",
   "job": "spreadsheets-data",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Here is my data as CSV: month,candles,soap,diffusers\n2026-05,410,320,95\n2026-06,395,340,120\n2026-07,380,365,150\n2026-08,420,390,170\n2026-09,455,410,210\n2026-10,470,430,260.\nI want to know: which product grew fastest from May to October 2026, in percent.\nCalculate the answer and show the calculation step by step. Then give the 3 most notable patterns with the numbers behind them.\nSay so if the data is too small to conclude anything.",
   "answer": "**Diffusers grew fastest, at about 173.7% from May to October 2026.**\n\n**Calculation** (percent change = (October − May) ÷ May × 100)\n\n| Product | May | Oct | Change | Calculation | % growth |\n|---|---|---|---|---|---|\n| Candles | 410 | 470 | +60 | 60 ÷ 410 × 100 | 14.6% |\n| Soap | 320 | 430 | +110 | 110 ÷ 320 × 100 | 34.4% |\n| Diffusers | 95 | 260 | +165 | 165 ÷ 95 × 100 | 173.7% |\n\n**Three notable patterns**\n\n1. **Diffusers grew every month and accelerated overall.** Month-over-month growth was +26.3% (95→120), +25.0% (120→150), +13.3% (150→170), +23.5% (170→210), and +23.8% (210→260). Their share of total units rose from 11.5% in May (95 of 825) to 22.4% in October (260 of 1,160).\n\n2. **Candles dipped before recovering.** Candles fell from 410 to 380 between May and July (−7.3%), then rose to 470 by October (+23.7% from July). Soap and diffusers did not show this dip.\n\n3. **Soap grew steadily.** Soap added between 20 and 25 units every month (+20, +25, +25, +20, +20), so its growth was consistent rather than volatile. It added the most absolute units (+110), but its percent growth was lower than diffusers' because it started from a larger base.\n\n**Caveat:** Six monthly data points covering one season is too little to conclude much. The diffusers' 173.7% figure is based on a small starting value (95), so a few extra units would change the percentage a lot. The patterns above describe this data, but they are not reliable predictions. Longer history would be needed to confirm any trend.",
   "format": "yes",
   "length": "n.a.",
   "invented": "minor",
   "missing": "n.a.",
   "result": "minor",
   "note": "Every figure checks out (candles +14.6%, soap +34.4%, diffusers +173.7%, shares 11.5% and 22.4%), but it called diffuser growth 'accelerated overall' when its monthly rates fell from 26% to 13% before rising again.",
   "seconds": 8.6,
   "cost_usd": 0.000795
  },
  {
   "case": "T16",
   "locale": "en",
   "prompt_id": "notes-to-actions",
   "job": "meetings-projects",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "From the meeting notes below, list:\n1) decisions made\n2) action items as a table: task, owner, due date\n3) open questions\nUse only names and dates that appear in the notes. Write \"no owner\" or \"no date\" where none was given.\n\nNotes:\nWebsite relaunch check-in, Oct 27\nPresent: Nadia, Omar, Jess\n- Agreed to launch on Nov 17, not Nov 10, because the product photos are late\n- Omar will get the final photos from the photographer by Nov 7\n- Jess to rewrite the shipping page, by Friday\n- Someone needs to test the checkout on phones\n- Nadia will ask the payment provider about Apple Pay\n- Decided to drop the blog section for launch\n- Open: do we redirect the old /shop URLs? Nobody knew",
   "answer": "## 1) Decisions made\n- Launch moved to **Nov 17** (not Nov 10), because the product photos are late.\n- **Drop the blog section** for launch.\n\n## 2) Action items\n\n| Task | Owner | Due date |\n|---|---|---|\n| Get final product photos from the photographer | Omar | Nov 7 |\n| Rewrite the shipping page | Jess | Friday |\n| Test checkout on phones | no owner | no date |\n| Ask the payment provider about Apple Pay | Nadia | no date |\n\n## 3) Open questions\n- Do we redirect the old /shop URLs? Nobody knew the answer.",
   "format": "yes",
   "length": "n.a.",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "Decisions, a 4-row action table and the open question. 'no owner' and 'no date' for the checkout test, 'no date' for Nadia.",
   "seconds": 4.4,
   "cost_usd": 0.000368
  },
  {
   "case": "T17",
   "locale": "en",
   "prompt_id": "status-update",
   "job": "meetings-projects",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a weekly status update on the new booking system for our three clinics for the clinic managers.\nThis week's raw notes: staff training done at Eastside and Midtown (24 people). Northgate training moved to next week, trainer sick. Data import of 6,200 patient records finished, 140 duplicates found, still being merged. Go-live planned Nov 3 but the SMS reminders are not working yet: the vendor says a fix by Oct 31, not confirmed. Need IT (Raj) to approve the firewall change for the SMS service. Next week: Northgate training, finish merging duplicates, test SMS\nFormat: overall status (on track, at risk or late) with one reason; done this week; next week; blockers, each with what I need from whom.\nUnder 150 words. Do not soften bad news.",
   "answer": "**Overall status: AT RISK.** Go-live (Nov 3) depends on SMS reminders, which are not working, and the vendor's Oct 31 fix is unconfirmed.\n\n**Done this week**\n- Staff training completed at Eastside and Midtown (24 staff).\n- Imported 6,200 patient records.\n- Found 140 duplicate records; merging in progress.\n\n**Next week**\n- Train Northgate staff (moved from this week after trainer illness).\n- Finish merging duplicates.\n- Test SMS reminders.\n\n**Blockers**\n- **SMS reminders not working.** Vendor says fix by Oct 31, not confirmed. *Needed from:* Vendor, confirmation of the fix date in writing.\n- **Firewall change for SMS service awaiting approval.** *Needed from:* Raj (IT), approval this week so SMS testing can happen next week.\n- **Northgate training delayed.** *Needed from:* Clinic managers, confirm Northgate staff availability for next week.",
   "format": "yes",
   "length": "yes",
   "invented": "minor",
   "missing": "n.a.",
   "result": "minor",
   "note": "122 words of 150, status 'at risk' with the right reason. Added a blocker the notes do not have: clinic managers to confirm Northgate staff availability.",
   "seconds": 6.6,
   "cost_usd": 0.00043
  },
  {
   "case": "T18",
   "locale": "en",
   "prompt_id": "job-description",
   "job": "hiring-hr",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a job description for a part-time bookkeeper at Harbor & Pine, an interior design studio in Savannah, Georgia, with 12 people.\nMust-haves: 2 years of bookkeeping for a small business; QuickBooks Online; payroll for up to 15 people; comfortable working on their own. Nice-to-haves: experience with project-based billing; Gusto. Salary range: $28 to $34 an hour. Location and hours: hybrid, 20 hours a week, two days a week in the Savannah studio.\nPlain, inclusive language, no \"rockstar\" or \"ninja\". Describe what the first 90 days look like.\nDo not add requirements I did not list.",
   "answer": "# Part-Time Bookkeeper\n\n**Harbor & Pine** · Savannah, Georgia\n**Interior design studio · 12 people**\n\n**Employment type:** Part-time, 20 hours per week\n**Schedule:** Hybrid. Two days a week in our Savannah studio; remaining hours remote.\n**Compensation:** $28 to $34 per hour, depending on experience\n\n---\n\n## About Harbor & Pine\n\nWe're a small interior design studio in Savannah. We work with homeowners, hospitality clients, and small businesses on projects that range from a single room to full renovations. Our team is small and close, and the finance work supports the creative work happening every day.\n\n## About the Role\n\nWe're looking for a part-time bookkeeper to keep our books accurate and up to date. You'll manage day-to-day transactions, support project billing, and run payroll for our team. You'll work mostly on your own, with regular check-ins with our leadership team.\n\n## What You'll Do\n\n- Record and categorize income and expenses in QuickBooks Online\n- Reconcile bank and credit card accounts each month\n- Prepare invoices and track payments for client projects\n- Run payroll for up to 15 employees and handle related tax filings and reporting\n- Prepare monthly financial reports for the leadership team\n- Keep organized records to support tax season and year-end close\n- Work with our outside accountant on questions and filings\n\n## What We're Looking For\n\n**Required**\n\n- Two years of bookkeeping experience for a small business\n- Hands-on experience with QuickBooks Online\n- Experience running payroll for up to 15 people\n- Comfortable working independently and managing your own priorities\n\n**Nice to have**\n\n- Experience with project-based billing, such as job costing or milestone invoicing\n- Experience with Gusto\n\n## Your First 90 Days\n\n**Days 1 to 30: Learn our systems**\n- Meet the team and learn how Harbor & Pine works day to day\n- Get set up in QuickBooks Online and review our chart of accounts\n- Shadow the current payroll cycle and learn our approval steps\n- Review open invoices and recent transactions with leadership\n- Set a regular weekly routine for reconciliations and billing\n\n**Days 31 to 60: Take ownership**\n- Run payroll on your own with a team member available to review\n- Complete a full month-end close, including bank and credit card reconciliation\n- Set up a process for tracking project billing and payments\n- Meet with leadership to identify any cleanup or process improvements\n\n**Days 61 to 90: Settle in**\n- Handle payroll, billing, and monthly reporting independently\n- Deliver your first monthly financial report to the leadership team\n- Agree on goals and a check-in schedule for the next six months\n- Share recommendations for improving our bookkeeping workflow\n\n---\n\n*Harbor & Pine is an equal opportunity employer. We welcome applicants from all backgrounds and encourage you to apply even if you don't meet every item on this list. If you need an accommodation during the application or interview process, please let us know.*",
   "format": "yes",
   "length": "n.a.",
   "invented": "major",
   "missing": "n.a.",
   "result": "fail",
   "note": "Requirements and salary exact, no 'rockstar', 90-day plan included, but it invented the studio's clients and projects, an outside accountant, tax filings and an equal opportunity statement. Root cause: the prompt had no slot for responsibilities.",
   "seconds": 8.6,
   "cost_usd": 0.00075
  },
  {
   "case": "T19",
   "locale": "en",
   "prompt_id": "interview-questions",
   "job": "hiring-hr",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write 8 interview questions for a customer support team lead that test handling escalations, coaching a team of 4, writing clear help articles.\nMix: 4 behavioral questions that ask for a real past example, 2 practical tasks, 2 about working with others.\nFor each, list what a strong answer includes and one red flag.\nNothing about age, family, religion, health or other personal characteristics.",
   "answer": "# Customer Support Team Lead Interview Questions\n\n## Behavioral Questions\n\n**1. Tell me about a time you handled an escalation that was going badly. What did you do, and what was the outcome?**\n- **Strong answer includes:** A specific case with a clear situation; actions taken to de-escalate (acknowledging the customer, owning next steps, involving the right people); how they kept the agent or team supported; a measurable or concrete result; a lesson they applied afterward.\n- **Red flag:** Describes the customer as the problem with no reflection on their own part, or can't name a specific case and speaks only in generalities.\n\n**2. Describe a time you coached a team member who was underperforming. What did you notice, and how did you approach the conversation?**\n- **Strong answer includes:** Specific observed behaviors rather than vague impressions; a private, direct conversation; questions to understand root cause (skills, tools, workload, confidence); an agreed plan with check-ins; follow-up that showed whether it worked.\n- **Red flag:** Jumps straight to discipline or performance warnings without describing coaching, or can't say what happened after the conversation.\n\n**3. Give an example of a help article or knowledge base content you wrote or significantly improved. How did you know it was working?**\n- **Strong answer includes:** Identification of the problem (repeat tickets, confusion, search failures); how they structured the content for the reader (task-first steps, plain language, screenshots where useful); how they tested it, such as with agents or a sample of customers; metrics like ticket deflection, reduced repeat contacts, or better CSAT on that topic.\n- **Red flag:** Describes the article only by its length or volume, with no evidence of reader outcomes, or says they never measured whether it helped.\n\n**4. Tell me about a time you had to change a process your team relied on. How did you bring people along?**\n- **Strong answer includes:** The reason for the change, grounded in data or customer impact; how they involved the team early and listened to objections; a transition plan with training or documentation; how they handled resistance and what they adjusted based on feedback.\n- **Red flag:** Says the change was imposed without discussion and the team just had to adapt, with no mention of how it was received or whether it was refined.\n\n## Practical Tasks\n\n**5. Walk me through how you would triage this queue: 40 open tickets, 6 flagged as escalations, two agents out sick, and a new product bug causing a spike in similar tickets. Talk through your first hour.**\n- **Strong answer includes:** Prioritizes by customer impact and SLA risk, not by who arrived first; identifies the bug as a likely common cause and considers a macro, status-page update, or engineering escalation to reduce duplicate work; reassigns or redistributes work realistically across available agents; communicates with stakeholders about expectations; plans a follow-up review once the immediate pressure eases.\n- **Red flag:** Works through tickets strictly in order without grouping the bug-related duplicates, or never mentions informing engineering or leadership about the spike.\n\n**6. Here's a help article draft that's too long and uses internal jargon: \"Upon successful authentication via the SSO gateway, users must navigate to the configuration panel to effect the requisite synchronization of their profile attributes...\" Rewrite the first two paragraphs for a non-technical customer, and tell me what you'd change about the structure.**\n- **Strong answer includes:** Plain, direct language (\"Sign in, then go to Settings and select Sync profile\"); short sentences; the most important action or answer first; a clear title and steps numbered in order; a note on what to do if the step fails; a suggestion to add a short intro stating who the article is for and what problem it solves; a plan to test with real customers or agents.\n- **Red flag:** Rewrites the text while keeping the jargon, or changes the wording without addressing structure, or ignores how the reader will find and use the article.\n\n## Working With Others\n\n**7. Tell me about a time you disagreed with a product, engineering, or leadership decision that affected your support team. How did you handle it?**\n- **Strong answer includes:** A clear articulation of the support impact and data behind their concern; a respectful approach to the other team, seeking to understand their constraints; a focus on shared goals; whether they committed to the decision after it was made, even if they didn't win the argument; practical steps taken to help the team handle the outcome.\n- **Red flag:** Describes the other team as unreasonable or uninformed, or says they simply complied without raising concerns or trying to influence the outcome.\n\n**8. How do you build a working relationship with a cross-functional partner, such as engineering or sales, who doesn't see support as a priority?**\n- **Strong answer includes:** Understanding what the partner cares about and speaking in their terms (deadlines, revenue, product quality); sharing support data proactively, such as recurring issues and their cost; regular touchpoints rather than only escalating when there's a crisis; recognizing their wins; a specific example of how the relationship produced a result.\n- **Red flag:** Treats the relationship as adversarial, or only contacts the partner when something is on fire, or assumes the other team should simply prioritize support without offering anything in return.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "8 questions in the 4, 2, 2 split, each with a strong answer and a red flag; nothing on personal characteristics.",
   "seconds": 15.0,
   "cost_usd": 0.001086
  },
  {
   "case": "T20",
   "locale": "en",
   "prompt_id": "resume-bullets",
   "job": "resume-career",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Rewrite these resume bullets for a operations manager application: - Was responsible for the warehouse team\n- Helped move inventory to a new system (we went from spreadsheets to NetSuite)\n- Cut shipping errors from about 4% to 1.5% over a year\n- Did the weekly schedule for 22 people\n- Worked with suppliers on delivery problems.\nEach bullet starts with a strong verb, stays under 25 words and shows a result.\nUse only numbers I gave. Where a number would help but I did not give one, write [ADD NUMBER: what to measure] instead of inventing it.",
   "answer": "1. Led a [ADD NUMBER: warehouse team headcount] person warehouse team to [ADD NUMBER: result, e.g., throughput or on-time shipping gain].\n2. Led migration of inventory records from spreadsheets to NetSuite, [ADD NUMBER: e.g., SKUs moved or hours saved per week].\n3. Cut shipping errors from about 4% to 1.5% over one year, improving order accuracy.\n4. Created weekly schedules for 22 staff, [ADD NUMBER: e.g., overtime hours reduced or shifts left uncovered].\n5. Resolved recurring supplier delivery problems, [ADD NUMBER: e.g., late deliveries reduced by X%].\n\nBullet 3 is the only one with a result you gave me. For the others, the placeholders ask for the figure that would make the bullet strongest. I didn't assume the warehouse team is the same 22 people you scheduled, so bullet 1 asks for its headcount.",
   "format": "yes",
   "length": "yes",
   "invented": "major",
   "missing": "yes",
   "result": "fail",
   "note": "All bullets under 25 words and [ADD NUMBER] used instead of invented metrics, but 'helped move inventory' became 'Led migration' and 'worked with suppliers' became 'Resolved recurring supplier delivery problems'.",
   "seconds": 13.0,
   "cost_usd": 0.001013
  },
  {
   "case": "T21",
   "locale": "en",
   "prompt_id": "cover-letter",
   "job": "resume-career",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Write a cover letter for Marketing Coordinator at Northwind Outdoor Gear.\nWhy this company, in my words: I've bought their repair kits for years and I like that they publish how to fix their gear instead of pushing new purchases.\nMy two most relevant achievements: grew our climbing gym's email list from 1,800 to 5,200 subscribers in 14 months; ran 9 community events in 2025 that brought in 340 new members.\nUnder 250 words in 3 paragraphs. No \"I am writing to apply\" opener and no clichés like \"passionate team player\".",
   "answer": "Dear Hiring Manager,\n\nI've bought Northwind repair kits for years, and the reason I keep coming back is that you publish how to fix your gear instead of steering people toward a replacement. I want to do marketing the same way: useful, honest content that earns trust before it asks for anything.\n\nAt our climbing gym, I grew the email list from 1,800 to 5,200 subscribers in 14 months. In 2025, I ran nine community events that brought in 340 new members. Those results taught me how to turn people who already care about the sport into a community that stays.\n\nI'd welcome the chance to bring that approach to Northwind's customers. Thank you for your time, and I'd be glad to talk about how I could help.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "128 words of 250 in 3 paragraphs, no stock opener or cliches, no facts about the company beyond the reason given.",
   "seconds": 7.4,
   "cost_usd": 0.000624
  },
  {
   "case": "T22",
   "locale": "en",
   "prompt_id": "explain-at-my-level",
   "job": "learning-explaining",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Explain the difference between cash flow and profit to me. What I already know: I run a small cleaning business and know my revenue and expenses, but not accounting terms. Why I need it: understand why my bank account is nearly empty when my accountant says I made a profit.\nStart with a 2-sentence answer, then one everyday analogy, then the 3 terms I will hear most, with one-line definitions.\nUnder 250 words. End with one question that checks whether I understood.",
   "answer": "Profit measures whether your business earns more than it spends over a period, counting sales and bills when they're earned or incurred. Cash flow measures when actual money moves in and out of your bank account. You can be profitable on paper and still run short of cash when the timing doesn't line up.\n\n**Analogy:** You cook and serve a big dinner for 10 people, but the guests pay you next month. You made a profit on the meal, yet your wallet is empty tonight because you still had to buy the groceries up front.\n\n**The 3 terms you'll hear most:**\n\n- **Accrual accounting:** Recording income and expenses when they happen, not when cash changes hands. This is why your accountant's profit number can look healthy while your bank balance doesn't.\n- **Accounts receivable:** Money customers owe you for work you've already done but haven't been paid for yet.\n- **Depreciation:** The yearly cost of equipment, such as vacuums or vans, spread across its useful life. It reduces profit but doesn't leave your bank account in that year.\n\n**Check your understanding:** If a client pays you for a job in March that you finished in February, which month does your accountant likely count the revenue in, and which month does your bank balance reflect it?",
   "format": "partial",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "minor",
   "note": "212 words of 250 with analogy, 3 terms and a closing question, but the opening answer was 3 sentences, not 2.",
   "seconds": 5.3,
   "cost_usd": 0.000618
  },
  {
   "case": "T23",
   "locale": "en",
   "prompt_id": "quiz-me",
   "job": "learning-explaining",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Quiz me on the material below. Ask 5 questions, one at a time, mixing recall and application.\nAfter each answer, tell me if it is right, explain in 2 sentences and quote the line of the material that answers it.\n\nMaterial:\nExpense policy, effective November 1, 2026\n- Meals while traveling: up to $60 a day, receipts required for anything over $25.\n- Hotels: book through the company portal; up to $210 a night in major cities and $150 elsewhere.\n- Mileage for personal cars: $0.70 a mile, logged in the expense app within 30 days.\n- Alcohol is never reimbursed.\n- Submit expenses within 30 days of the trip; late claims need a manager's approval.\n- Client gifts over $75 need approval before purchase.",
   "answer": "Question 1 of 5:\n\nUnder this policy, what is the daily cap on meal reimbursement while traveling, and above what dollar amount do you need a receipt?",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Asked question 1 of 5 and waited, without answering it.",
   "seconds": 2.2,
   "cost_usd": 0.000287
  },
  {
   "case": "T24",
   "locale": "en",
   "prompt_id": "say-no-politely",
   "job": "writing-email",
   "round": "r1",
   "prompt_version": "v1",
   "prompt": "Help me say no to speaking for free at the Riverside Business Association breakfast on December 9 from Carla, the association's events chair.\nMy real reason, in a few words: I am fully booked with client work until January.\nWhat I can offer instead: a 2-page checklist on hiring a first employee that they can share with members.\nWrite it as a email, warm but firm, under 100 words.\nMake the no clear in the first two sentences. Apologize at most once, and do not hint at a yes I have not offered.",
   "answer": "Subject: Re: Riverside Business Association breakfast, December 9\n\nHi Carla,\n\nThank you for thinking of me for the December 9 breakfast. I have to decline. I'm fully booked with client work through January, so I can't take on speaking commitments right now.\n\nI'd still like to support the members. I can offer a two-page checklist on hiring your first employee, which you're welcome to share with the group.\n\nWould that be useful? I'm happy to send it over.\n\nBest,\n[Your name]",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "81 words of 100, the no in sentence 2, no apology, only the checklist offered.",
   "seconds": 4.3,
   "cost_usd": 0.000321
  },
  {
   "case": "T03",
   "locale": "en",
   "prompt_id": "product-description",
   "job": "marketing-social",
   "round": "r2",
   "prompt_version": "v2",
   "prompt": "Write a product description for the Fjord Wool Throw blanket for my Shopify product page.\nFacts: 100% merino wool, 130 x 170 cm, $89, machine washable on the wool cycle at 30 C, made in Portugal, comes rolled in a cotton bag, colors oat, slate and moss\nIdeal buyer: people furnishing a small apartment who want one warm, good-looking throw for the sofa\nFormat: a benefit headline of one line, one paragraph of 3 or 4 short sentences (80 words at most), then 4 bullets that each pair a feature with what it means for the buyer.\nDo not invent certifications, awards, reviews, origins or measurements.",
   "answer": "# Fjord Wool Throw in Oat, Slate and Moss\n\n**Wrap your sofa in warmth that looks as good as it feels.**\n\nMade from 100% merino wool, the Fjord Wool Throw brings soft, cozy comfort to a small space without crowding it. Its generous 130 x 170 cm size covers you on the couch or sits neatly over an armchair. Choose from three calm colors, oat, slate or moss, to suit your room. It arrives rolled in a cotton bag, ready to gift or store.\n\n- **100% merino wool:** Soft, warm and breathable, so you stay cozy without overheating on long evenings.\n- **130 x 170 cm:** Big enough to share with your legs and feet, yet compact enough for a small apartment.\n- **Machine washable on the wool cycle at 30 °C:** Keeps it fresh without a trip to the dry cleaner.\n- **Made in Portugal, in oat, slate and moss:** Three versatile neutrals that pair easily with your existing decor.",
   "format": "yes",
   "length": "yes",
   "invented": "minor",
   "missing": "n.a.",
   "result": "minor",
   "note": "Paragraph 63 words in 4 sentences. Added 'breathable' and 'without overheating' to the merino, which the facts do not say.",
   "seconds": 6.3,
   "cost_usd": 0.000425
  },
  {
   "case": "T03",
   "locale": "en",
   "prompt_id": "product-description",
   "job": "marketing-social",
   "round": "r3",
   "prompt_version": "v2",
   "prompt": "Write a product description for the Fjord Wool Throw blanket for my Shopify product page.\nFacts: 100% merino wool, 130 x 170 cm, $89, machine washable on the wool cycle at 30 C, made in Portugal, comes rolled in a cotton bag, colors oat, slate and moss\nIdeal buyer: people furnishing a small apartment who want one warm, good-looking throw for the sofa\nFormat: a benefit headline of one line, one paragraph of 3 or 4 short sentences (80 words at most), then 4 bullets that each pair a feature with what it means for the buyer.\nDo not invent certifications, awards, reviews, origins or measurements.",
   "answer": "# Fjord Wool Throw\n\n**Wrap up in warmth that fits your small space.**\n\nOur Fjord Wool Throw is made from 100% merino wool and measures 130 x 170 cm, a generous size for the sofa or armchair. It's a soft, good-looking layer that adds warmth without crowding a compact living room. Available in oat, slate and moss, it's made in Portugal and comes rolled in a cotton bag, ready to gift or store.\n\n- **100% merino wool:** Soft, warm and naturally breathable, so it keeps you cozy without overheating.\n- **130 x 170 cm:** Big enough to cover your lap and shoulders, yet compact enough to suit a small apartment.\n- **Machine washable on the wool cycle at 30°C:** Easy care that keeps the throw looking fresh without a trip to the dry cleaner.\n- **Oat, slate and moss colors, plus a cotton bag:** Pick a shade that complements your sofa, and the included bag keeps it tidy when not in use.",
   "format": "yes",
   "length": "yes",
   "invented": "minor",
   "missing": "n.a.",
   "result": "minor",
   "note": "Paragraph 60 words in 3 sentences. Added 'naturally breathable' again.",
   "seconds": 6.1,
   "cost_usd": 0.00035
  },
  {
   "case": "T05",
   "locale": "en",
   "prompt_id": "cold-email",
   "job": "sales",
   "round": "r2",
   "prompt_version": "v2",
   "prompt": "Write a cold email to a office manager at a dental practice with 2 to 5 dentists.\nWhat I sell: a text-message service that confirms appointments 48 hours ahead. The problem it solves for them: no-shows and last-minute cancellations that leave chairs empty.\nProof I can mention: none yet. The ask: a 15-minute call next week.\nDescribe the offer only as I wrote it: no extra features, statistics or claims about their industry.\nUnder 120 words, plain text, no flattering opener and no fake familiarity.\nLeave [PERSONAL LINE] where I will add one real detail about them. Give two subject lines.",
   "answer": "**Subject line options:**\n\n1. Empty chairs from missed appointments\n2. Cutting last-minute cancellations at [Practice Name]\n\n**Email:**\n\nHi [First Name],\n\n[PERSONAL LINE]\n\nI run a text-message service that confirms patient appointments 48 hours before they're scheduled. It's built for practices dealing with no-shows and last-minute cancellations that leave chairs empty.\n\nWould you be open to a 15-minute call next week? I can walk you through how it works and you can tell me whether it fits your office.\n\nBest,\n[Your Name]\n[Phone or Email]",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "84 words, the offer described only as given, [PERSONAL LINE] kept.",
   "seconds": 5.6,
   "cost_usd": 0.000349
  },
  {
   "case": "T05",
   "locale": "en",
   "prompt_id": "cold-email",
   "job": "sales",
   "round": "r3",
   "prompt_version": "v2",
   "prompt": "Write a cold email to a office manager at a dental practice with 2 to 5 dentists.\nWhat I sell: a text-message service that confirms appointments 48 hours ahead. The problem it solves for them: no-shows and last-minute cancellations that leave chairs empty.\nProof I can mention: none yet. The ask: a 15-minute call next week.\nDescribe the offer only as I wrote it: no extra features, statistics or claims about their industry.\nUnder 120 words, plain text, no flattering opener and no fake familiarity.\nLeave [PERSONAL LINE] where I will add one real detail about them. Give two subject lines.",
   "answer": "Subject lines:\n\n1. Fewer empty chairs at [Practice Name]\n2. Appointment confirmations 48 hours ahead\n\nEmail:\n\nHi [First Name],\n\n[PERSONAL LINE]\n\nI run a text-message service that confirms patient appointments 48 hours before they happen. It's built for practices dealing with no-shows and last-minute cancellations that leave chairs empty.\n\nWould you be open to a 15-minute call next week to see if it fits [Practice Name]?\n\nThanks,\n[Your Name]\n[Phone or Email]",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "72 words, the offer described only as given, [PERSONAL LINE] kept.",
   "seconds": 5.3,
   "cost_usd": 0.00027
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r2",
   "prompt_version": "v2",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe.\nOffer nothing beyond the remedy above, and add no promises about outcomes, such as what a visit will fix.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Dear [Customer name],\n\nThank you for telling us about the three outages this month. Losing a full day of client calls on Monday is serious, especially when you work from home. You've been with us for six years, and you should have heard back from us after your last ticket.\n\nHere's what we can do. We'll credit three days of service to your next bill. We'll also send a technician to your home this Thursday between 8 am and noon, at no charge to you. Please reply to this email or call us to confirm the time works for you.\n\nI'm not able to offer the $400 you asked for, but the credit and visit above are what we can provide.\n\nWarm regards,\n[Your name]\nBrightwave Internet",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "127 words, only the credit and the Thursday visit, declined the $400 plainly.",
   "seconds": 10.5,
   "cost_usd": 0.000888
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r3",
   "prompt_version": "v2",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe.\nOffer nothing beyond the remedy above, and add no promises about outcomes, such as what a visit will fix.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Hi [Customer name],\n\nI'm sorry this is the third outage this month, and I understand how much Monday's lost client calls cost you. You've been with us six years, and you should not have had to chase us after your last ticket.\n\nI can't offer $400 in compensation, but here is what I can do. We'll apply a credit for three days of service to your next bill, and a technician will visit this Thursday between 8 am and noon at no charge, since the fault is on our side. Please reply to confirm the time works, or call us at [phone number] if you need a different slot, and we'll confirm the booking within one business day.\n\nIf you'd like to talk through the account history or your options, reply here and a member of our team will follow up with you directly.\n\n[Your name]\nBrightwave Internet",
   "format": "yes",
   "length": "no",
   "invented": "major",
   "missing": "n.a.",
   "result": "fail",
   "note": "148 words (limit 140). Said the fault was on our side, promised to confirm the booking within one business day and a follow-up from the team.",
   "seconds": 6.8,
   "cost_usd": 0.0003
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r2",
   "prompt_version": "v2",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nGive: one title under 60 characters with the keyword near the start, a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].\nDo not state fees, statistics or rules of thumb as facts: write them as [VERIFY: the claim] so I can check them.",
   "answer": "# How to Price Handmade Jewelry: A Formula That Covers Materials, Time, and Fees\n\n**Direct answer (for the top of the article):**\nPrice each handmade jewelry piece by adding up materials, the labor you pay yourself, and the platform and payment fees, then applying a markup that covers overhead and profit. Check the result against what similar pieces sell for, and adjust so the price is both profitable and competitive. [MY PROOF: one real piece you've made, with its actual materials cost, time, and final price, to use as a worked example]\n\n---\n\n## What Should I Include When Calculating the Cost of a Piece?\n\n- Define \"materials cost\" to include beads, findings, wire, chain, clasps, packaging, and tags, not just the main stones or charms. [VERIFY: recommended approach for counting small consumables like jump rings and polishing cloths]\n- Explain how to calculate cost per unit when you buy in bulk (cost of package ÷ number of usable units). [MY PROOF: screenshot of a supplier receipt showing bulk cost and quantity]\n- Note that waste and breakage should be factored in. [VERIFY: typical waste percentage for beadwork and metalwork]\n- Give a sample materials breakdown for one earring pair and one necklace. [MY PROOF: your own parts list with real prices]\n\n## How Do I Value My Time?\n\n- Explain the difference between making time and admin time (photos, listings, shipping, customer messages). [MY PROOF: a rough log of how long one listing and one order take you]\n- Show how to set an hourly rate and multiply it by the time per piece. [VERIFY: current hourly rate range for skilled handmade artisans]\n- Address the trap of pricing at \"minimum wage\" and why it leads to burnout. [MY PROOF: a personal experience or reader-style example of underpricing]\n\n## Which Fees Should I Subtract From My Price?\n\n- List the common platform fees on Etsy: listing fees, transaction fees, payment processing, and offsite ads, with a note that they change. [VERIFY: current Etsy listing fee, transaction fee, and payment processing rates]\n- Explain craft fair costs: booth fees, table rental, card reader fees, travel, and sales tax handling. [VERIFY: typical craft fair booth fee ranges and card reader fee percentages]\n- Show how to build fees into the price so they are not taken from profit. [MY PROOF: screenshot of your Etsy fee breakdown from one real sale]\n\n## How Do I Add Profit and Overhead to My Price?\n\n- Explain overhead: tools, equipment, workspace, software, and insurance, spread across expected sales. [VERIFY: common overhead allocation methods for small makers]\n- Introduce a markup or profit margin and how it is applied on top of cost. [VERIFY: typical markup multiplier used by handmade jewelry sellers, such as the \"2x to 3x materials\" rule and its limits]\n- Warn that markup rules of thumb are starting points, not fixed rules. [MY PROOF: a before-and-after example of a piece you repriced]\n\n## How Do I Check If My Price Will Actually Sell?\n\n- Explain how to research competitors: search similar pieces, filter by sold items on Etsy, and compare materials and quality. [MY PROOF: screenshots of three comparable listings and their prices]\n- Discuss adjusting for your brand level (beginner, mid-range, premium) and the buyer you're targeting. [VERIFY: typical price bands for handmade jewelry by category]\n- Explain how to test a price by tracking views, favorites, and sales over a few weeks. [MY PROOF: your own sales data from a listing you changed price on]\n\n## What Are Common Pricing Mistakes to Avoid?\n\n- Forgetting to include fees or packaging costs. [MY PROOF: an example of a sale where you forgot a cost and lost money]\n- Pricing too low to compete with mass-produced jewelry, and how to communicate value instead. [VERIFY: examples or data on buyer willingness to pay for handmade versus mass-made]\n- Failing to raise prices as materials costs change. [MY PROOF: a supplier price increase you experienced and how you handled it]\n- Ignoring wholesale and discount pricing if you sell to shops or run sales. [VERIFY: standard wholesale discount percentage for handmade jewelry]\n\n## How Do I Put the Pricing Formula Together?\n\n- Walk through a full formula step by step: materials + time + fees + overhead + profit = price. [MY PROOF: a complete worked calculation for one real piece]\n- Provide a simple spreadsheet or template reader can copy. [MY PROOF: screenshot of your own pricing sheet]\n- End with a reminder to review prices every few months. [VERIFY: suggested review frequency]",
   "format": "partial",
   "length": "no",
   "invented": "none",
   "missing": "n.a.",
   "result": "fail",
   "note": "[VERIFY] marks on every fee and rule of thumb, nothing stated as fact, but the only title was 78 characters.",
   "seconds": 19.4,
   "cost_usd": 0.001054
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r3",
   "prompt_version": "v2",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nGive: one title under 60 characters with the keyword near the start, a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].\nDo not state fees, statistics or rules of thumb as facts: write them as [VERIFY: the claim] so I can check them.",
   "answer": "# Title\n**How to Price Handmade Jewelry: A Step-by-Step Formula** (52 characters)\n\n# Direct Answer (for the top of the article)\nTo price handmade jewelry, add up the materials, your labor time at an hourly rate you choose, and the platform and payment fees for each piece, then add a profit margin. Compare the total with what similar pieces sell for in your market, and adjust your labor rate or margin until the price covers your costs and still sells.\n\n# Outline\n\n## What costs go into the price of a handmade piece?\n- List the four cost groups: materials, labor, overhead, and fees\n- Explain why forgetting packaging, tools, and display costs leads to underpricing\n- [MY PROOF] A photo of your own piece with its materials laid out\n\n## How do I calculate the material cost of one piece?\n- Show how to divide a bulk purchase (for example, a strand of beads) into the cost per piece\n- Explain how to count waste, such as clasps you lose or wire you cut short\n- [MY PROOF] Your real receipt or supplier invoice, with the per-piece math worked out\n\n## How do I value my time?\n- Explain how to choose an hourly rate, and why it should cover more than the time at the bench (design, photos, listing, shipping, and admin)\n- Give a method for timing one piece from start to finish\n- [MY PROOF] Your own time log for one piece, or a screenshot of your timer app\n- [VERIFY: typical hourly rates for handmade jewelry makers, if you cite any figure]\n\n## Which fees should I include?\n- Separate platform listing fees, transaction fees, and payment processing fees\n- Explain craft fair costs such as table or booth fees, and how to spread them across the pieces you expect to sell\n- Note that fees change, so readers should check the current schedule\n- [VERIFY: current Etsy listing, transaction, and payment processing fees and the date you checked them]\n- [VERIFY: current craft fair booth fee ranges, if you cite them]\n\n## How much profit should I add on top?\n- Explain profit as the money that keeps the business running and pays for growth, not as an extra\n- Show how to add a margin to the subtotal of materials, labor, overhead, and fees\n- [VERIFY: any \"standard\" markup or margin you suggest, such as a multiplier rule of thumb]\n- [MY PROOF] A worked example of your margin and the reasoning behind it\n\n## How do I check my price against the market?\n- Explain how to find comparable pieces on Etsy and at local fairs by material, size, and craftsmanship\n- Warn against matching the lowest price, since it may not cover your costs\n- [MY PROOF] Screenshots of three to five comparable listings, with your notes on where your piece sits\n\n## Should I price differently for Etsy and craft fairs?\n- Explain how fees, shipping, and in-person sales change the final price\n- Cover how wholesale prices for shops differ from retail prices\n- [MY PROOF] Your own Etsy price next to your fair price for the same piece\n\n## How do I know whether my price is working?\n- Describe what to track: sales per listing, views, conversion, and returns or price objections\n- Explain when to raise prices, and how to do it without losing repeat buyers\n- [MY PROOF] A sales summary from your shop or fair records, with one adjustment you made\n\n## Worked example: pricing one necklace from start to finish\n- Walk through each cost group with real numbers for one piece\n- Show the final price and the check against the market\n- [MY PROOF] Your actual necklace or bracelet, its full cost breakdown, and its final listing price",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Title 53 characters, [VERIFY] on every fee and markup rule.",
   "seconds": 14.2,
   "cost_usd": 0.001117
  },
  {
   "case": "T18",
   "locale": "en",
   "prompt_id": "job-description",
   "job": "hiring-hr",
   "round": "r2",
   "prompt_version": "v2",
   "prompt": "Write a job description for a part-time bookkeeper at Harbor & Pine, an interior design studio in Savannah, Georgia, with 12 people.\nMain responsibilities: monthly bookkeeping and bank reconciliations in QuickBooks Online, invoicing clients and tracking payments for design projects, running payroll twice a month, a one-page monthly report for the two partners\nMust-haves: 2 years of bookkeeping for a small business; QuickBooks Online; payroll for up to 15 people; comfortable working on their own. Nice-to-haves: experience with project-based billing; Gusto. Salary range: $28 to $34 an hour. Location and hours: hybrid, 20 hours a week, two days a week in the Savannah studio.\nPlain, inclusive language, no \"rockstar\" or \"ninja\". Describe what the first 90 days look like.\nDo not add requirements, duties, benefits or facts about the company that I did not give. Where a section needs more, write [ADD: what is missing].",
   "answer": "# Part-Time Bookkeeper\n\n**Harbor & Pine** · Savannah, Georgia · Hybrid · 20 hours per week\n**Compensation:** $28-$34 per hour\n\n## About Harbor & Pine\n\nHarbor & Pine is an interior design studio in Savannah, Georgia. We are a team of 12, led by two partners. [ADD: a few sentences on what the studio does and the kind of projects it takes on]\n\n## The Role\n\nWe are looking for a part-time bookkeeper to keep our finances accurate and organized. You will manage monthly books and bank reconciliations, handle client invoicing and payment tracking for design projects, and run payroll twice a month. You will also prepare a one-page monthly report for the two partners. You will work mostly on your own, with support from the partners.\n\n## What You Will Do\n\n- Complete monthly bookkeeping and bank reconciliations in QuickBooks Online\n- Create client invoices and track payments for design projects\n- Run payroll twice a month\n- Prepare a one-page monthly financial report for the two partners\n\n## What We Are Looking For\n\n**Must-have**\n- Two years of bookkeeping experience for a small business\n- Experience using QuickBooks Online\n- Experience running payroll for up to 15 people\n- Comfort working independently\n\n**Nice-to-have**\n- Experience with project-based billing\n- Experience using Gusto\n\n## Location and Hours\n\nThis role is hybrid. You will work 20 hours per week, with two days in our Savannah studio and the rest remote. [ADD: which days are in-studio and whether remote days are flexible]\n\n## Benefits\n\n[ADD: benefits, or confirm there are none]\n\n## Your First 90 Days\n\n**Days 1-30: Learn our systems.** Get set up in QuickBooks Online and our payroll tool, and review how invoicing and project billing currently work. Complete your first full month of bank reconciliations and your first two payroll runs with support available. [ADD: who you will meet with and how onboarding is structured]\n\n**Days 31-60: Work independently.** Take over the monthly close and invoicing for design projects. Run payroll on your own, with a partner reviewing the first cycle. Begin to note any process questions or gaps you find. [ADD: check-in schedule with the partners]\n\n**Days 61-90: Own the full cycle.** Complete a full monthly close and deliver the one-page report to the partners. Review the process with them and agree on any changes to how the books, invoicing, or payroll are handled. [ADD: how success will be measured at the 90-day mark]\n\n## How to Apply\n\n[ADD: application instructions, deadline, and contact person]",
   "format": "yes",
   "length": "n.a.",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Responsibilities exactly as given, [ADD: ...] for the studio description, benefits, days and how to apply. Wrote '$28-$34' with an en dash.",
   "seconds": 11.4,
   "cost_usd": 0.000931
  },
  {
   "case": "T18",
   "locale": "en",
   "prompt_id": "job-description",
   "job": "hiring-hr",
   "round": "r3",
   "prompt_version": "v2",
   "prompt": "Write a job description for a part-time bookkeeper at Harbor & Pine, an interior design studio in Savannah, Georgia, with 12 people.\nMain responsibilities: monthly bookkeeping and bank reconciliations in QuickBooks Online, invoicing clients and tracking payments for design projects, running payroll twice a month, a one-page monthly report for the two partners\nMust-haves: 2 years of bookkeeping for a small business; QuickBooks Online; payroll for up to 15 people; comfortable working on their own. Nice-to-haves: experience with project-based billing; Gusto. Salary range: $28 to $34 an hour. Location and hours: hybrid, 20 hours a week, two days a week in the Savannah studio.\nPlain, inclusive language, no \"rockstar\" or \"ninja\". Describe what the first 90 days look like.\nDo not add requirements, duties, benefits or facts about the company that I did not give. Where a section needs more, write [ADD: what is missing].",
   "answer": "# Part-Time Bookkeeper\n\n**Harbor & Pine** · Savannah, Georgia\n**Employment type:** Part-time, 20 hours per week\n**Location:** Hybrid. Two days per week in our Savannah studio. [ADD: which two days, or whether they are set or flexible]\n**Pay:** $28 to $34 per hour\n**Start date:** [ADD: target start date]\n**Reports to:** The two partners\n\n---\n\n## About Harbor & Pine\n\nHarbor & Pine is an interior design studio in Savannah, Georgia. We have 12 people on our team. [ADD: a short description of the studio's work, clients, and values]\n\n## About the Role\n\nWe are looking for a part-time bookkeeper to keep our financial records accurate and up to date. You will handle monthly bookkeeping, track invoices and payments for design projects, and run payroll. You will also prepare a short monthly report for the two partners. You will work mostly on your own, with the partners available for questions.\n\n## Responsibilities\n\n- Complete monthly bookkeeping and bank reconciliations in QuickBooks Online\n- Create invoices for design clients and track project payments\n- Run payroll twice a month\n- Prepare a one-page monthly financial report for the two partners\n\n## Requirements\n\n**Must have:**\n- At least 2 years of bookkeeping experience for a small business\n- Experience using QuickBooks Online\n- Experience running payroll for up to 15 people\n- Comfortable working independently\n\n**Nice to have:**\n- Experience with project-based billing\n- Experience with Gusto\n\n## Compensation and Benefits\n\n- Pay: $28 to $34 per hour, depending on experience\n- Benefits: [ADD: benefits offered, such as paid time off, holidays, retirement plan, or health coverage, or state that none are offered]\n\n## How to Apply\n\n[ADD: application instructions, such as how to apply, what to include (resume, cover letter, references), and the deadline]\n\nHarbor & Pine is an equal opportunity employer. [ADD: equal opportunity statement or accommodation language, if the studio uses one]\n\n---\n\n## What Your First 90 Days Look Like\n\n**Days 1 to 30: Learn our systems**\n- Get access to QuickBooks Online and our bank accounts, and review how the books are currently set up\n- Meet with the partners to learn how we handle client invoices, project payments, and payroll\n- Complete your first bank reconciliation with a partner or with the partners' guidance\n- Prepare your first monthly report and review it with the partners\n- Run payroll with support, and confirm each step with the partners before it is final\n\n**Days 31 to 60: Work on your own**\n- Complete monthly bookkeeping and reconciliations on your own schedule\n- Run payroll twice a month and handle the process independently\n- Set up and track invoices for active design projects\n- Check in with the partners as questions come up, and keep a list of anything that needs a decision\n\n**Days 61 to 90: Own the process**\n- Complete a full monthly close and report without needing a review of each step\n- Review the process for invoicing and payment tracking, and suggest changes to the partners\n- Confirm with the partners how the monthly report should look going forward\n- Meet with the partners for a 90-day check-in to talk about what is working and what to adjust\n\n[ADD: any formal review schedule, training resources, or milestones the studio uses]",
   "format": "yes",
   "length": "n.a.",
   "invented": "major",
   "missing": "n.a.",
   "result": "fail",
   "note": "[ADD: ...] marks used, but it still wrote 'Harbor & Pine is an equal opportunity employer'.",
   "seconds": 13.9,
   "cost_usd": 0.000883
  },
  {
   "case": "T20",
   "locale": "en",
   "prompt_id": "resume-bullets",
   "job": "resume-career",
   "round": "r2",
   "prompt_version": "v2",
   "prompt": "Rewrite these resume bullets for a operations manager application: - Was responsible for the warehouse team\n- Helped move inventory to a new system (we went from spreadsheets to NetSuite)\n- Cut shipping errors from about 4% to 1.5% over a year\n- Did the weekly schedule for 22 people\n- Worked with suppliers on delivery problems.\nEach bullet starts with a strong verb, stays under 25 words and shows a result.\nKeep my level of responsibility: if I wrote \"helped\", do not write \"led\".\nUse only numbers I gave. Where a number would help but I did not give one, write [ADD NUMBER: what to measure] instead of inventing it.",
   "answer": "1. Oversaw daily warehouse operations for the [ADD NUMBER: team size] people on the warehouse team.\n2. Supported the move of inventory records from spreadsheets to NetSuite, covering [ADD NUMBER: SKUs or inventory items migrated].\n3. Cut shipping errors from about 4% to 1.5% over a year.\n4. Built the weekly schedule for 22 people, keeping [ADD NUMBER: shift coverage rate or hours saved per week].\n5. Coordinated with suppliers to resolve delivery problems, reducing [ADD NUMBER: late or missed deliveries] by [ADD NUMBER: percent].\n\nI used \"Oversaw,\" \"Supported,\" and \"Coordinated\" to match the level of responsibility you described. Bullet 3 is the only one with a result you gave me in full. If you can supply the placeholder numbers, they will make bullets 1, 2, 4, and 5 stronger.",
   "format": "yes",
   "length": "yes",
   "invented": "minor",
   "missing": "yes",
   "result": "minor",
   "note": "Kept 'Supported' for 'helped' and 'Coordinated' for 'worked with'. Bullet 5 still implies a reduction ('reducing [ADD NUMBER] by [ADD NUMBER]').",
   "seconds": 12.2,
   "cost_usd": 0.00114
  },
  {
   "case": "T20",
   "locale": "en",
   "prompt_id": "resume-bullets",
   "job": "resume-career",
   "round": "r3",
   "prompt_version": "v2",
   "prompt": "Rewrite these resume bullets for a operations manager application: - Was responsible for the warehouse team\n- Helped move inventory to a new system (we went from spreadsheets to NetSuite)\n- Cut shipping errors from about 4% to 1.5% over a year\n- Did the weekly schedule for 22 people\n- Worked with suppliers on delivery problems.\nEach bullet starts with a strong verb, stays under 25 words and shows a result.\nKeep my level of responsibility: if I wrote \"helped\", do not write \"led\".\nUse only numbers I gave. Where a number would help but I did not give one, write [ADD NUMBER: what to measure] instead of inventing it.",
   "answer": "1. Managed the warehouse team of [ADD NUMBER: team size], keeping operations on schedule and [ADD NUMBER: a measurable outcome, such as on-time shipments or safety incidents].\n2. Supported the migration of inventory from spreadsheets to NetSuite, [ADD NUMBER: e.g., SKUs or locations moved, or weeks to go live].\n3. Reduced shipping errors from about 4% to 1.5% over one year.\n4. Built the weekly schedule for a 22-person team, [ADD NUMBER: e.g., overtime hours saved or coverage rate].\n5. Coordinated with suppliers to address delivery problems, [ADD NUMBER: e.g., late deliveries reduced or number of suppliers managed].\n\nNotes on the choices:\n- \"Managed\" matches \"was responsible for.\" I didn't use \"Led,\" since you didn't describe leading people beyond scheduling them.\n- \"Supported\" is a little stronger than \"helped\" but keeps your role as a contributor to the migration.\n- \"Coordinated\" is used for the supplier work because you said you worked with suppliers, not that you resolved the problems. If you did resolve them, change it to \"Resolved.\"\n- Bullets 1, 2, 4, and 5 are each under 25 words once the placeholders are filled in with short numbers.",
   "format": "yes",
   "length": "yes",
   "invented": "minor",
   "missing": "yes",
   "result": "minor",
   "note": "Kept the level of responsibility and explained each verb. Bullet 1 adds 'keeping operations on schedule'.",
   "seconds": 11.8,
   "cost_usd": 0.000921
  },
  {
   "case": "T22",
   "locale": "en",
   "prompt_id": "explain-at-my-level",
   "job": "learning-explaining",
   "round": "r2",
   "prompt_version": "v2",
   "prompt": "Explain the difference between cash flow and profit to me. What I already know: I run a small cleaning business and know my revenue and expenses, but not accounting terms. Why I need it: understand why my bank account is nearly empty when my accountant says I made a profit.\nOpen with the answer in exactly 2 sentences, then one everyday analogy, then the 3 terms I will hear most, with one-line definitions.\nUnder 250 words. End with one question that checks whether I understood.",
   "answer": "Profit measures whether your business earns more than it spends over a period, while cash flow measures when money actually arrives in and leaves your bank account. You can show a profit on paper and still run short on cash, because customers pay late, you buy supplies up front, and some costs (like equipment) are spread out over time in the books but paid all at once.\n\n**Analogy:** Think of a cleaning crew that finishes a $1,000 job on the 1st but gets paid on the 60th. The job counts as earned income right away, yet for two months you still had to pay wages and buy supplies from your own pocket.\n\n**The 3 terms you'll hear most:**\n\n- **Revenue:** The total money your business earns from sales, whether or not it's been paid yet.\n- **Accounts receivable:** Money customers owe you for work you've already done but haven't paid.\n- **Net income (profit):** Revenue minus all expenses for a period, as calculated on the books.\n\n**Quick check:** If you invoice a client $2,000 in March but they pay in May, does your March profit go up or down, and does your March bank balance go up or down?",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Answer in 2 sentences, 196 words, analogy, 3 terms, closing question.",
   "seconds": 3.8,
   "cost_usd": 0.000416
  },
  {
   "case": "T22",
   "locale": "en",
   "prompt_id": "explain-at-my-level",
   "job": "learning-explaining",
   "round": "r3",
   "prompt_version": "v2",
   "prompt": "Explain the difference between cash flow and profit to me. What I already know: I run a small cleaning business and know my revenue and expenses, but not accounting terms. Why I need it: understand why my bank account is nearly empty when my accountant says I made a profit.\nOpen with the answer in exactly 2 sentences, then one everyday analogy, then the 3 terms I will hear most, with one-line definitions.\nUnder 250 words. End with one question that checks whether I understood.",
   "answer": "Profit is what's left after subtracting your expenses from your revenue, counting work when you do it, whether or not the money has arrived yet. Cash flow is the money actually moving into and out of your bank account, so a business can show a profit while its balance is nearly empty.\n\n**Analogy:** You clean an office and invoice the client, who pays in 60 days. The job counts as profit right away, but your bank balance doesn't change until they pay, and you still have to pay your supplies and staff in the meantime.\n\n**The 3 terms you'll hear most:**\n\n- **Net profit:** Your total revenue minus all your expenses for a period, based on when work was done and billed.\n- **Cash flow:** The actual money coming in and going out of your bank account over a period.\n- **Accounts receivable:** Money customers owe you for work you've already completed but haven't paid yet.\n\nYour accountant's profit figure likely includes unpaid invoices, so some of it may still be sitting with clients rather than in your account.\n\n**Quick check:** If you finish a $1,000 job in March but the client pays in May, which month shows the $1,000 as profit, and which month shows it in your bank account?",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Answer in 2 sentences, 208 words, analogy, 3 terms, closing question.",
   "seconds": 5.8,
   "cost_usd": 0.000583
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r4",
   "prompt_version": "v3",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe.\nUse only the policy and remedy above: no other commitments, such as callbacks, follow-ups, causes or what a visit will fix.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Dear [Customer name],\n\nThree outages in one month is unacceptable, and losing a full day of client calls on Monday is a real cost to you. You've been with us for six years, and no one called you back after your last ticket. That should not have happened.\n\nHere is what I can do now. I've added a credit for three days of service to your next bill. I've also booked a free technician visit for this Thursday, between 8 am and noon, at no charge to you.\n\nI know this doesn't cover your lost income, and I understand if it doesn't settle things for you.\n\n[Your name]\nBrightwave Internet",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "110 words, only the remedy, no cause and no extra promise.",
   "seconds": 9.4,
   "cost_usd": 0.000826
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r5",
   "prompt_version": "v3",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe.\nUse only the policy and remedy above: no other commitments, such as callbacks, follow-ups, causes or what a visit will fix.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Here's a reply you can send:\n\n---\n\nHi,\n\nYou're right to be frustrated. Three outages this month, a lost day of client calls on Monday, and no callback after your last ticket is not the service you've paid six years for.\n\nFor the outage, we'll credit three days of service on your next bill, and a technician will visit this Thursday between 8 am and noon to look at the fault, which we'll cover at no charge since it's on our side.\n\nI can't approve the $400 for lost income, but the credit and visit above are what I can commit to right now. If you'd like to talk through the lost work, reply here and a member of our team will review your account with you.\n\nThank you for staying with us,\n[Name]\nBrightwave Internet\n\n---\n\nWord count: about 120 words.",
   "format": "yes",
   "length": "yes",
   "invented": "major",
   "missing": "n.a.",
   "result": "fail",
   "note": "Reply about 117 words, but said the fault was on our side and that 'a member of our team will review your account'.",
   "seconds": 5.2,
   "cost_usd": 0.000299
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r4",
   "prompt_version": "v3",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nStart with the title alone on the first line: under 60 characters, keyword near the start. Then give a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].\nDo not state fees, statistics or rules of thumb as facts: write them as [VERIFY: the claim] so I can check them.",
   "answer": "How to Price Handmade Jewelry: A Simple Formula That Covers Costs and Sells\n\nThe direct answer: Price each piece by adding up what it costs you to make (materials, your hourly labor, and a share of overhead), then adding the platform and payment fees you'll pay when it sells. Then check that the total is something buyers in your niche will pay, and adjust the labor rate or the design, not just the number, until the margin and the market both work.\n\n## What does a handmade jewelry piece actually cost to make?\n\n- List every material per piece: beads, wire, clasps, chain, findings, packaging. Use a per-unit cost, not what you paid for the bulk pack.\n- Show how to convert bulk purchases into per-piece cost (e.g., $ paid ÷ number of usable units).\n- Include waste and breakage allowance and how you set it.\n- [MY PROOF]: A real bill of materials for one of your best-selling pieces, with your actual supplier receipts.\n- [MY PROOF]: A screenshot of your supplier invoice or spreadsheet showing per-unit calculation.\n\n## How do I value my time when pricing jewelry?\n\n- Explain timing each step (designing, cutting, assembling, finishing, photographing, packing) with a stopwatch or timer over several pieces.\n- Set an hourly rate you want to earn, and show how to multiply it by the minutes per piece.\n- Explain why you should not price labor at zero, and how that causes burnout.\n- [VERIFY: typical hourly rates for handmade jewelry makers in your country or marketplace]\n- [MY PROOF]: Your own timed log of how long a specific piece takes, from start to finished and packed.\n\n## What overhead should I include in my jewelry price?\n\n- Define overhead: tools, workspace, electricity, packaging, business software, booth fees for craft fairs.\n- Explain how to allocate monthly overhead across expected pieces sold per month.\n- Show a simple monthly overhead spreadsheet example.\n- [VERIFY: whether overhead should be allocated per piece or added as a percentage in your market]\n- [MY PROOF]: Your actual monthly overhead costs from the last 3 months.\n\n## What fees will I pay on Etsy and at craft fairs?\n\n- Break down Etsy listing fees, transaction fees, payment processing, and offsite ads fees, with a note to check current rates on the Etsy seller fee page.\n- Break down craft fair costs: booth or table fees, card reader fees, sales tax, and travel.\n- Show how to build fees into the price, including how to gross up a price so fees don't eat your profit.\n- [VERIFY: current Etsy listing fee, transaction fee, payment processing rate, and offsite ads rate]\n- [VERIFY: typical craft fair booth fee ranges and card reader fees]\n- [MY PROOF]: A screenshot of your most recent Etsy payout statement showing the fees deducted from one sale.\n\n## How do I set a profit margin for handmade jewelry?\n\n- Explain how to add a profit margin on top of costs, not just cover them.\n- Compare cost-plus pricing with a target margin approach and explain when each is useful.\n- Show a worked example: materials + labor + overhead + fees = total cost, then add margin to reach a selling price.\n- [VERIFY: common profit margin targets for handmade or small-batch goods]\n- [MY PROOF]: A worked calculation for one piece, from your real numbers, ending with the final listing price.\n\n## How do I check that my price will still sell?\n\n- Explain how to compare your price against similar pieces on Etsy, using search results and sold listings filters.\n- Explain how to check craft fair competitors' prices and what buyers expect at your event.\n- Suggest a test: list at the calculated price, track views, favorites, and sales for 4 to 6 weeks, then adjust.\n- Explain when to lower cost (cheaper materials, simpler design) rather than lower price.\n- [VERIFY: whether Etsy's sold-listings filter and search results are still the best way to see comparable sales]\n- [MY PROOF]: Screenshots of three competing listings for similar pieces, with their prices and your notes on how yours compare.\n\n## What mistakes should I avoid when pricing jewelry?\n\n- Pricing by copying competitors without checking your own costs.\n- Forgetting to include packaging, shipping materials, or fees.\n- Discounting too often, which trains buyers to wait for sales.\n- Raising prices without changing anything about the piece or your brand.\n- [MY PROOF]: An example of a pricing mistake you made and what you changed.",
   "format": "yes",
   "length": "no",
   "invented": "none",
   "missing": "n.a.",
   "result": "fail",
   "note": "Title alone on the first line as asked, but 75 characters. [VERIFY] held.",
   "seconds": 12.5,
   "cost_usd": 0.001016
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r5",
   "prompt_version": "v3",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nStart with the title alone on the first line: under 60 characters, keyword near the start. Then give a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].\nDo not state fees, statistics or rules of thumb as facts: write them as [VERIFY: the claim] so I can check them.",
   "answer": "How to Price Handmade Jewelry So It Covers Costs and Still Sells\n\nPricing handmade jewelry means adding up every cost behind a piece (materials, your hourly time, platform and payment fees, packaging, and a profit margin) and then checking that the total fits what buyers in your market will pay. Use a simple formula, price each piece individually, and revisit the numbers whenever your costs change.\n\n## What costs should I include when pricing a handmade jewelry piece?\n\n- List every material per piece: beads, wire, findings, clasps, chain, stones, and glue, with the cost per unit and how much one piece uses.\n- Include a share of consumables and tools that wear out (pliers, polish, sandpaper, tumbling media) as a small per-piece amount.\n- Add packaging: box or pouch, card, tissue, and a thank-you insert.\n- Mention waste or breakage: a set percentage for lost beads or failed pieces. [VERIFY: a typical waste allowance for jewelry making and whether it is worth tracking separately]\n- [MY PROOF]: Your own material cost sheet for one real best-selling piece, with actual supplier receipts.\n\n## How do I value my time when pricing handmade jewelry?\n\n- Time each step (cutting, assembling, finishing, photographing, listing, packing) for at least five pieces, then average it.\n- Choose an hourly rate you want to earn and multiply it by the average minutes per piece.\n- Explain why unpaid design and admin time still counts, and how to decide whether to include it. [VERIFY: current guidance on whether handmade sellers should bill for admin and listing time]\n- Warn about underpricing: pay yourself at least a basic wage, or the business will not last. Keep the advice general unless you have a figure to back it up.\n- [MY PROOF]: A timing log from one of your own pieces, showing the minutes you actually spent.\n\n## What fees should I subtract when pricing on Etsy and at craft fairs?\n\n- Explain the main Etsy charges: listing fee, transaction fee, payment processing fee, and any offsite or advertising fees. [VERIFY: current Etsy listing fee, transaction fee, and payment processing rates, in the seller's currency and region]\n- Explain craft fair costs: booth or table fee, travel, parking, display, card reader fees, and sales tax handling. [VERIFY: typical craft fair booth fee ranges and card reader fees]\n- Show how to build fees into the price: divide the selling price by one minus the total percentage fee, then add any fixed per-sale fee. [VERIFY: confirm the formula and whether fixed fees apply per order or per item]\n- [MY PROOF]: A screenshot of your own Etsy fee breakdown from a recent sale, and your craft fair expense receipts.\n\n## How do I add a profit margin and set the final price?\n\n- Add the subtotal of materials, labor, and fees, then add a profit margin on top.\n- Explain that the margin is what funds growth, slow seasons, and taxes, not just a gift to yourself. [VERIFY: a commonly recommended profit margin range for handmade goods]\n- Show a worked example from start to finish with a single piece.\n- Round to a clean price point that is easy to remember and easy for buyers to compare.\n- [MY PROOF]: A worked example using one of your actual pieces, with the final price you listed.\n\n## How do I check that my price will still sell?\n\n- Compare your price against similar pieces from other makers on Etsy and at fairs, looking at the same materials and quality level.\n- Check whether buyers in your niche are price-sensitive or willing to pay for craftsmanship, branding, or story.\n- Test a price change on a few listings or at a fair before changing the whole line, and track views, favorites, and sales.\n- Explain when to raise prices: when demand is steady, when costs rise, or when a piece sells out quickly. [VERIFY: whether sell-out speed is a reliable signal to raise prices]\n- [MY PROOF]: Screenshots of your Etsy shop stats or sales log before and after a price change, or notes from a craft fair where you tested two price points.\n\n## How do I keep my pricing up to date?\n\n- Recalculate your cost sheet every time a supplier price changes or you switch materials.\n- Review fee changes from Etsy and your payment processor at least once a year. [VERIFY: how often Etsy announces fee changes]\n- Keep a simple spreadsheet with cost, time, fees, and final price per piece so you can spot problems quickly.\n- Explain how to apply price increases to existing stock and new listings, and how to notify repeat buyers.\n- [MY PROOF]: A copy of your pricing spreadsheet template or a screenshot of your own version.",
   "format": "yes",
   "length": "no",
   "invented": "none",
   "missing": "n.a.",
   "result": "fail",
   "note": "Title 63 characters. [VERIFY] held.",
   "seconds": 13.4,
   "cost_usd": 0.00096
  },
  {
   "case": "T18",
   "locale": "en",
   "prompt_id": "job-description",
   "job": "hiring-hr",
   "round": "r4",
   "prompt_version": "v3",
   "prompt": "Write a job description for a part-time bookkeeper at Harbor & Pine, an interior design studio in Savannah, Georgia, with 12 people.\nMain responsibilities: monthly bookkeeping and bank reconciliations in QuickBooks Online, invoicing clients and tracking payments for design projects, running payroll twice a month, a one-page monthly report for the two partners\nMust-haves: 2 years of bookkeeping for a small business; QuickBooks Online; payroll for up to 15 people; comfortable working on their own. Nice-to-haves: experience with project-based billing; Gusto. Salary range: $28 to $34 an hour. Location and hours: hybrid, 20 hours a week, two days a week in the Savannah studio.\nPlain, inclusive language, no \"rockstar\" or \"ninja\". Describe what the first 90 days look like.\nDo not add requirements, duties, benefits, legal statements or facts about the company that I did not give. Where a section needs more, write [ADD: what is missing].",
   "answer": "# Part-Time Bookkeeper\n\n**Harbor & Pine** | Savannah, Georgia | Hybrid | 20 hours per week | $28 to $34 per hour\n\n## About Harbor & Pine\n\nHarbor & Pine is an interior design studio in Savannah, Georgia. We have 12 people on our team and are led by two partners. [ADD: a short description of the studio's work, clients, and values]\n\n## The Role\n\nWe are hiring a part-time bookkeeper to keep our financial records accurate and up to date. You will handle monthly bookkeeping, track payments on design projects, run payroll, and give our two partners a clear monthly summary of where the business stands.\n\n## What You Will Do\n\n- Complete monthly bookkeeping and bank reconciliations in QuickBooks Online\n- Create client invoices for design projects and track payments until they are received\n- Run payroll twice a month for the team\n- Prepare a one-page monthly report for the two partners\n\n## What We Are Looking For\n\n**Required**\n- At least 2 years of bookkeeping experience for a small business\n- Experience using QuickBooks Online\n- Experience running payroll for up to 15 people\n- Comfort working independently and managing your own schedule\n\n**Nice to Have**\n- Experience with project-based billing\n- Experience with Gusto\n\n## Compensation\n\n$28 to $34 per hour, depending on experience. [ADD: whether there is any paid time off, holiday pay, or other benefits]\n\n## Location and Hours\n\n- 20 hours per week\n- Hybrid: two days per week at our Savannah studio, with the remaining hours [ADD: where the other hours are worked]\n- [ADD: set working days or flexible schedule, and typical start and end times]\n\n## Your First 90 Days\n\n**Days 1 to 30: Learn our systems**\n- Get access to QuickBooks Online and our bank accounts\n- Review the last several months of bookkeeping and reconciliations with a partner to understand how we currently record transactions\n- Learn how our invoicing and project payment tracking works\n- Meet the partners and agree on the format and timing of the monthly report [ADD: who you will meet and how often]\n\n**Days 31 to 60: Take on the monthly cycle**\n- Complete one full monthly close and bank reconciliation\n- Send invoices and track open payments for active projects\n- Run your first payroll cycles, checking each one with a partner before it is final [ADD: payroll provider currently used and who approves payroll]\n\n**Days 61 to 90: Work independently**\n- Own the monthly close and reconciliation from start to finish\n- Deliver the monthly report to the partners on schedule\n- Review the process with the partners and note any changes to improve accuracy or save time [ADD: check-in schedule and what success looks like at day 90]\n\n## How to Apply\n\n[ADD: application instructions, deadline, contact person, and whether a resume, cover letter, or work samples are needed]",
   "format": "yes",
   "length": "n.a.",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "No company facts, legal lines or duties beyond the input; [ADD: ...] where sections needed more.",
   "seconds": 11.7,
   "cost_usd": 0.001068
  },
  {
   "case": "T18",
   "locale": "en",
   "prompt_id": "job-description",
   "job": "hiring-hr",
   "round": "r5",
   "prompt_version": "v3",
   "prompt": "Write a job description for a part-time bookkeeper at Harbor & Pine, an interior design studio in Savannah, Georgia, with 12 people.\nMain responsibilities: monthly bookkeeping and bank reconciliations in QuickBooks Online, invoicing clients and tracking payments for design projects, running payroll twice a month, a one-page monthly report for the two partners\nMust-haves: 2 years of bookkeeping for a small business; QuickBooks Online; payroll for up to 15 people; comfortable working on their own. Nice-to-haves: experience with project-based billing; Gusto. Salary range: $28 to $34 an hour. Location and hours: hybrid, 20 hours a week, two days a week in the Savannah studio.\nPlain, inclusive language, no \"rockstar\" or \"ninja\". Describe what the first 90 days look like.\nDo not add requirements, duties, benefits, legal statements or facts about the company that I did not give. Where a section needs more, write [ADD: what is missing].",
   "answer": "# Part-Time Bookkeeper\n\n**Harbor & Pine** · Savannah, Georgia\n**Type:** Part-time, 20 hours per week\n**Location:** Hybrid. Two days a week in our Savannah studio. [ADD: where the remaining work hours happen]\n**Pay:** $28 to $34 per hour\n\n## About Harbor & Pine\n\nHarbor & Pine is an interior design studio in Savannah, Georgia. We are a team of 12. [ADD: a few sentences about the studio's work, clients, and way of working]\n\n## The Role\n\nWe are looking for a bookkeeper to keep our financial records accurate and up to date. You will work closely with the two partners, who will rely on your monthly numbers to make decisions about the studio. [ADD: who the bookkeeper reports to day to day, and who they work with on the team]\n\n## What You'll Do\n\n- Complete monthly bookkeeping and bank reconciliations in QuickBooks Online\n- Create client invoices for design projects, and track payments until they are received\n- Run payroll twice a month for our team\n- Prepare a one-page monthly report for the two partners\n\n## What We're Looking For\n\n**Required**\n- Two years of bookkeeping experience for a small business\n- Experience using QuickBooks Online\n- Experience running payroll for up to 15 people\n- Comfort working on your own and managing your own time\n\n**Nice to have**\n- Experience with project-based billing\n- Experience with Gusto\n\n## Your First 90 Days\n\n**Days 1 to 30: Learn the systems.**\nIn your first month, you will get access to QuickBooks Online, the bank accounts, and the payroll setup. You will meet the two partners and talk through how they use the monthly report and what they need from it. You will complete one full month of bookkeeping and bank reconciliation with support available, and review the process with a partner at the end.\n\n**Days 31 to 60: Take on the full cycle.**\nYou will run payroll on your own, twice a month, and take over client invoicing and payment tracking for design projects. You will complete the monthly close and reconciliations independently, and check in with the partners on any open questions.\n\n**Days 61 to 90: Own the work.**\nBy the end of your third month, you will be handling the full set of responsibilities on your own. You will deliver the monthly one-page report to the partners without edits to the core numbers. You will also review with the partners what is working and what could be improved in the process. [ADD: any specific goals or milestones the partners want to set for this period]\n\n## How to Apply\n\n[ADD: how to apply, application deadline, and what to include]\n\n[ADD: start date]",
   "format": "yes",
   "length": "n.a.",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "No company facts, legal lines or duties beyond the input; [ADD: ...] where sections needed more.",
   "seconds": 10.3,
   "cost_usd": 0.000799
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r6",
   "prompt_version": "v4",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe.\nUse only the policy and remedy above: no other commitments, such as callbacks, follow-ups, causes or what a visit will fix.\nBefore you finish, delete any sentence that commits us to something not in the remedy.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Hi [Name],\n\nThree outages this month, including the full day of client calls you lost on Monday, is not acceptable for someone who works from home. You also should have heard back after your last ticket.\n\nHere's what we can offer:\n\n- A credit for three days of service on your next bill.\n- A free technician visit this Thursday, between 8 am and noon.\n\nI can't promise the $400 you asked for, but these are the steps I can put in place for you now.\n\n[Your name]\nBrightwave Internet\n\n---\n\nI removed every sentence that committed to anything beyond the remedy. The $400 line declines that request without promising anything, so check that you're comfortable with that wording before sending.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "118 words with a note to us; only the remedy, the $400 declined.",
   "seconds": 10.8,
   "cost_usd": 0.000953
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r7",
   "prompt_version": "v4",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe.\nUse only the policy and remedy above: no other commitments, such as callbacks, follow-ups, causes or what a visit will fix.\nBefore you finish, delete any sentence that commits us to something not in the remedy.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Dear [Name],\n\nThree outages this month, including a full day of client calls lost on Monday, is a serious disruption for someone who works from home. After six years as a customer, you should not have had to chase us after your last ticket, and I'm sorry that happened.\n\nHere is what we're offering. Your next bill will carry a credit for three days of service. We'll also send a technician this Thursday between 8 am and noon, at no charge to you.\n\nSincerely,\n[Name]\nBrightwave Internet",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "87 words, only the remedy.",
   "seconds": 10.0,
   "cost_usd": 0.000649
  },
  {
   "case": "T07",
   "locale": "en",
   "prompt_id": "complaint-reply",
   "job": "customer-support",
   "round": "r8",
   "prompt_version": "v4",
   "prompt": "Write a reply to this customer complaint for Brightwave Internet, a regional internet provider.\nOur policy: outages longer than 24 hours earn a credit of one day of service per day affected; technician visits are free when the fault is on our side. What I can offer: a credit for 3 days of service on the next bill and a free technician visit this Thursday between 8 am and noon.\nWarm, accountable and specific, without corporate phrases like \"we apologize for any inconvenience\". Under 140 words.\nName the specific problem they describe, state the fix and the next step with a timeframe.\nUse only the policy and remedy above: no other commitments, such as callbacks, follow-ups, causes or what a visit will fix.\nBefore you finish, delete any sentence that commits us to something not in the remedy.\n\nComplaint:\nThis is the third time this month my internet has gone down. I work from home and lost a full day of client calls on Monday. I've been a customer for six years and nobody called me back after my last ticket. I want compensation for my lost income, around $400, or I'm switching providers.",
   "answer": "Dear [Name],\n\nThree outages this month, including the full day of client calls you lost on Monday, is not acceptable, especially after six years with us and the follow-up you had to make after your last ticket.\n\nHere's what we're doing. Your next bill will carry a credit for three days of service. Because the fault is on our side, the technician visit is free, and it's set for this Thursday between 8 am and noon.\n\n[Sign-off]\n\n*Note: The $400 request for lost income isn't covered by our policy, so the reply doesn't address it. You may want to decide separately how to respond to that part.*",
   "format": "yes",
   "length": "yes",
   "invented": "major",
   "missing": "n.a.",
   "result": "fail",
   "note": "No extra promise, but it stated the cause: 'Because the fault is on our side'.",
   "seconds": 11.7,
   "cost_usd": 0.000804
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r6",
   "prompt_version": "v4",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nStart with 3 title options, each under 60 characters with the keyword near the start and its character count in brackets. Then give a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].\nDo not state fees, statistics or rules of thumb as facts: write them as [VERIFY: the claim] so I can check them.",
   "answer": "**Title options**\n\n1. How to Price Handmade Jewelry: A Simple Formula [47 chars]\n2. How to Price Handmade Jewelry Without Underselling [50 chars]\n3. Handmade Jewelry Pricing: Cover Costs and Still Sell [52 chars]\n\n**Direct answer**\n\nPrice each piece by adding up materials, the time you spend making and listing it, packaging, and the fees each sale costs you, then adding a margin you can live with. Check that number against what similar pieces sell for, and adjust if it's too far above or below the market.\n\n## What does a handmade piece actually cost me to make?\n\n- Materials per piece: beads, findings, wire, chain, clasps, priced by the unit you use [MY PROOF: receipts or a cost sheet for one real piece]\n- Waste and breakage allowance [VERIFY: typical waste percentage to add to materials]\n- Tools and workspace costs spread across pieces, and how you estimate that share\n\n## How do I put a value on my time?\n\n- Time each step: design, cutting, assembly, finishing [MY PROOF: timer log for one real piece]\n- Choose an hourly rate you would accept for this work [VERIFY: current hourly rates for handmade jewelers on Etsy and at fairs]\n- Count admin time too: photos, listing, packing, customer messages\n\n## What fees come out of each sale?\n\n- Marketplace listing and transaction fees, payment processing, and any offsite ad costs [VERIFY: current Etsy fee schedule and payment processing rates]\n- Craft fair costs: booth or table fee, card reader fees, travel, display [MY PROOF: your actual fair invoice and receipts]\n- Shipping and packaging: whether you absorb them or pass them on\n- Refund and return allowance [VERIFY: typical refund rate for handmade jewelry shops]\n\n## How do I build the final price from these numbers?\n\n- Worked example adding materials, time, packaging, and fees, then margin [MY PROOF: one worked example using your real numbers]\n- Rounding to a clean price point\n- A copyable spreadsheet template with the formula laid out [MY PROOF: screenshot of your own spreadsheet]\n\n## Will the price still sell?\n\n- Compare against 3 to 5 listings for similar pieces [MY PROOF: screenshots of comparable listings and where yours sits among them]\n- Wholesale versus retail pricing, and why they differ [VERIFY: common wholesale discount ranges for makers]\n- Test and adjust: track views, favorites, and sales for a few weeks after a change [MY PROOF: your sales data before and after one price change]\n- Signs you're priced too low (buyers assume it's cheap or ask for discounts) versus too high (few views, no sales)\n\n## How do I raise prices without losing customers?\n\n- Raising prices on new listings first, then existing ones\n- Wording a price change announcement for repeat buyers [MY PROOF: an example announcement you've sent or would send]\n- When to hold off on a change, such as during a sale or after a shop rule change [VERIFY: current Etsy policy on price changes for existing orders]",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Three titles of 47 to 52 characters, counts exact; 6 [VERIFY] and 8 [MY PROOF] marks; no number stated as fact.",
   "seconds": 15.8,
   "cost_usd": 0.001563
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r7",
   "prompt_version": "v4",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nStart with 3 title options, each under 60 characters with the keyword near the start and its character count in brackets. Then give a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].\nDo not state fees, statistics or rules of thumb as facts: write them as [VERIFY: the claim] so I can check them.",
   "answer": "**Titles**\n\n1. How to Price Handmade Jewelry: A Cost-Based Guide [49]\n2. How to Price Handmade Jewelry for Etsy and Craft Fairs [54]\n3. How to Price Handmade Jewelry So It Actually Sells [50]\n\n**Direct answer**\n\nA handmade jewelry price has to cover materials, the time you spend making each piece, platform and payment fees, and a share of your overhead, with some left over as profit. Once the cost floor is set, compare the price against what similar pieces sell for so the number is both sustainable and competitive. [VERIFY: confirm the cost categories you want to include, and whether your readers should treat this as a floor or a target]\n\n---\n\n## What costs belong in the price of a handmade piece?\n\n- Name the four cost buckets: materials, labor, fees, overhead.\n- Explain why a price that only covers materials leaves the maker losing money on every sale.\n- [MY PROOF] A breakdown of one real piece you made, with every cost listed.\n\n## How do I calculate the materials cost for each piece?\n\n- Show how to find the cost per unit for beads, wire, clasps, chain, and findings when bought in bulk.\n- Explain how to split bulk purchases across pieces and how to count waste or breakage.\n- [VERIFY: current bulk pricing examples from your main suppliers]\n- [MY PROOF] Screenshot of a supplier receipt and your per-piece materials calculation.\n\n## How do I value my time without underpricing my work?\n\n- Explain how to time a piece from start to finish, including design, assembly, finishing, and packing.\n- Offer a method for choosing an hourly rate and show the math.\n- Point out that time spent on admin, photos, and shipping is also labor.\n- [VERIFY: typical hourly rates for handmade jewelry makers in your market]\n- [MY PROOF] Your own timing log for two or three pieces.\n\n## What fees will Etsy and craft fairs take from each sale?\n\n- List the fee types a seller typically pays: listing, transaction, payment processing, and any others that apply.\n- Explain that fees are charged on the total the buyer pays, including shipping, when that applies.\n- Note that craft fair fees (booth, table, card reader) work differently and should be spread across expected sales.\n- [VERIFY: current Etsy listing, transaction, and payment processing fees, and current craft fair booth costs]\n- [MY PROOF] A screenshot of your actual Etsy fee breakdown for one recent order.\n\n## How much overhead and profit should I add?\n\n- Define overhead: tools, packaging, studio space, software, website, and insurance.\n- Show how to spread monthly overhead across an expected number of pieces sold.\n- Explain profit as the money that funds growth, replacing tools, and future time off.\n- [VERIFY: any commonly cited markup or profit percentage, presented as a range with its source]\n- [MY PROOF] Your monthly overhead list and how many pieces you actually sell each month.\n\n## How do I check that my price will sell?\n\n- Compare your price against similar pieces: same material, similar size, similar design level.\n- Explain how to read the market by looking at sold listings, not only active ones.\n- Cover when a higher price signals quality and when it drives buyers away.\n- [VERIFY: how to access sold-listing data on Etsy and any current limitations]\n- [MY PROOF] Screenshots of three competing listings and what you noticed about their prices.\n\n## Should my Etsy price and craft fair price be different?\n\n- Explain why fair sales avoid online fees but add booth and travel costs.\n- Discuss pricing for wholesale or bundle deals, and how to avoid undercutting your own online shop.\n- [MY PROOF] A comparison of your Etsy and fair prices for one piece, with the reasons for each.\n\n## How do I review and adjust prices over time?\n\n- Recommend a review schedule tied to material cost changes, sales data, and fee changes.\n- Explain how to raise prices on existing pieces without losing repeat buyers.\n- Give a simple checklist readers can use each review.\n- [VERIFY: any platform policies on price changes or sale-price display]\n- [MY PROOF] A before-and-after example of a price you raised and what happened to sales.",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "pass",
   "note": "Three titles of 49 to 54 characters, counts exact; 7 [VERIFY] marks.",
   "seconds": 16.4,
   "cost_usd": 0.001463
  },
  {
   "case": "T10",
   "locale": "en",
   "prompt_id": "article-outline",
   "job": "seo-content",
   "round": "r8",
   "prompt_version": "v4",
   "prompt": "Outline an article for the search \"how to price handmade jewelry\". Readers: people who sell handmade jewelry on Etsy or at craft fairs. What they should be able to do after reading: set a price for each piece that covers materials, time and fees and still sells.\nStart with 3 title options, each under 60 characters with the keyword near the start and its character count in brackets. Then give a direct answer of 2 sentences for the top, H2 headings phrased like real questions, and under each H2 the points to cover.\nMark the points where I need my own data, screenshots or examples with [MY PROOF].\nDo not state fees, statistics or rules of thumb as facts: write them as [VERIFY: the claim] so I can check them.",
   "answer": "# Title options\n\n1. **How to Price Handmade Jewelry: A Step-by-Step Guide** [49 chars]\n2. **Pricing Handmade Jewelry So You Profit (Not Just Break Even)** [63 chars]: too long, so I'll use this one instead:\n3. **Handmade Jewelry Pricing: Cover Costs, Time, and Fees** [56 chars]\n\nOption 2 is over 60 characters and doesn't start with the keyword, so here is a corrected version:\n2. **How to Price Handmade Jewelry Without Underselling Yourself** [60 chars]: at the limit, so I'll shorten it: **How to Price Handmade Jewelry Without Underselling** [48 chars]\n\nFinal list:\n1. **How to Price Handmade Jewelry: A Step-by-Step Guide** [49 chars]\n2. **How to Price Handmade Jewelry Without Underselling** [48 chars]\n3. **Handmade Jewelry Pricing: Cover Costs, Time, and Fees** [56 chars]\n\n# Direct answer\n\nTo price handmade jewelry, add up the cost of materials, the wages you want for your time, and the marketplace and payment fees on each sale, then set a price that covers all three and still fits what buyers pay for similar pieces. Test that price against competitor listings and your own sales data, and adjust it as your costs change.\n\n## What do I need to count as the cost of each piece?\n\n- Materials per piece: beads, wire, chain, clasps, findings, packaging. Break bulk purchases down to a cost per unit. [MY PROOF: a receipt or spreadsheet showing your actual cost for one finished piece]\n- Waste and breakage allowance: how much you lose to mistakes or damaged stock. [VERIFY: a common waste allowance percentage for jewelry makers]\n- Tools and equipment spread across pieces: pliers, torches, tumblers, and how you divide their cost over expected output. [MY PROOF: your tool list with purchase prices]\n- Overhead such as studio space, electricity, and software, allocated per piece. [VERIFY: how to allocate overhead per unit for a home-based craft business]\n\n## How do I calculate my labor and time?\n\n- Time each step separately: cutting, assembling, finishing, photographing, and packing. [MY PROOF: a timed log for one piece]\n- Choose an hourly wage you want to earn and multiply it by the time per piece. [VERIFY: typical hourly rates handmade sellers use, and how to justify your figure]\n- Note that time for admin like customer messages and shipping counts too. [MY PROOF: an example of an order's total admin time]\n\n## What fees come out of every sale?\n\n- Marketplace listing and transaction fees on Etsy. [VERIFY: current Etsy listing, transaction, and payment processing fees and any changes in 2026]\n- Payment processing fees for craft fair card readers and payment apps. [VERIFY: current typical card reader and payment app fee rates]\n- Shipping costs and packaging, and whether you pass them to buyers. [MY PROOF: a screenshot of your shipping charges and actual postage cost]\n- Sales tax handling, which depends on your location. [VERIFY: sales tax rules for online and in-person craft sales in your state or country]\n\n## How do I put the numbers together into one price?\n\n- Use the formula: materials + labor + overhead + fees = minimum price. [VERIFY: the formula is a standard cost-plus approach; confirm the fee deduction method you use]\n- Add a profit margin on top. [VERIFY: a healthy profit margin range for handmade goods]\n- Show a worked example with one real necklace or earring set from start to finish. [MY PROOF: a complete costing worksheet for one of your pieces]\n\n## How do I check that the price will actually sell?\n\n- Compare your price with similar pieces on Etsy and at craft fairs, noting materials and craftsmanship. [MY PROOF: screenshots of three comparable listings and what you noticed about them]\n- Look at your own sales data to see which prices convert and which sit unsold. [MY PROOF: a screenshot of your Etsy stats or a sales log]\n- Test a higher price on one piece or collection and track views and sales. [MY PROOF: a before-and-after example from your shop, if you have one]\n\n## When should I review and change my prices?\n\n- Revisit prices when material costs change or fees update. [VERIFY: how often fee schedules change on Etsy and payment platforms]\n- Raise prices gradually and tell regular customers when you do. [MY PROOF: an example of a price change announcement you've used or would use]\n- Keep a simple pricing log with the date and reason for each change. [MY PROOF: a sample pricing log template]",
   "format": "partial",
   "length": "yes",
   "invented": "none",
   "missing": "n.a.",
   "result": "minor",
   "note": "The final three titles are 48 to 56 characters, but it showed its corrections on the page and two counts were off by 2 to 3.",
   "seconds": 12.3,
   "cost_usd": 0.000967
  },
  {
   "case": "T16es",
   "locale": "es",
   "prompt_id": "notes-to-actions",
   "job": "meetings-projects",
   "round": "t1",
   "prompt_version": "translated",
   "prompt": "De las notas de reunión de abajo, haz una lista de:\n1) decisiones tomadas\n2) tareas en una tabla: tarea, responsable, fecha límite\n3) preguntas abiertas\nUsa solo nombres y fechas que aparezcan en las notas. Escribe \"sin responsable\" o \"sin fecha\" donde no se haya dado ninguno.\n\nNotas:\nReunión de relanzamiento de la tienda en línea, 27 de octubre\nPresentes: Lucía, Andrés, Paula\n- Acordamos lanzar el 17 de noviembre y no el 10, porque las fotos de producto llegan tarde\n- Andrés conseguirá las fotos finales del fotógrafo antes del 7 de noviembre\n- Paula reescribe la página de envíos, para el viernes\n- Alguien tiene que probar el pago en celulares\n- Lucía le va a preguntar al proveedor de pagos por Mercado Pago\n- Decidimos quitar el blog para el lanzamiento\n- Pendiente: ¿redirigimos las URL viejas de /tienda? Nadie sabía",
   "answer": "# Reunión de relanzamiento de la tienda en línea (27 de octubre)\n\n## 1) Decisiones tomadas\n- Lanzar el **17 de noviembre**, no el 10, porque las fotos de producto llegan tarde.\n- **Quitar el blog** para el lanzamiento.\n\n## 2) Tareas\n\n| Tarea | Responsable | Fecha límite |\n|---|---|---|\n| Conseguir las fotos finales del fotógrafo | Andrés | 7 de noviembre |\n| Reescribir la página de envíos | Paula | viernes (sin fecha exacta) |\n| Probar el pago en celulares | sin responsable | sin fecha |\n| Preguntar al proveedor de pagos por Mercado Pago | Lucía | sin fecha |\n\n## 3) Preguntas abiertas\n- ¿Redirigimos las URL viejas de /tienda? Nadie sabía la respuesta.",
   "format": "yes",
   "length": "n.a.",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "Spanish: decisions, a 4-row table and the open question; 'sin responsable' and 'sin fecha' for the checkout test, 'sin fecha' for Lucia.",
   "seconds": 5.3,
   "cost_usd": 0.000445
  },
  {
   "case": "T08es",
   "locale": "es",
   "prompt_id": "faq-from-tickets",
   "job": "customer-support",
   "round": "t1",
   "prompt_version": "translated",
   "prompt": "Estas son preguntas recientes de clientes: 1. ¿Puedo probar una clase antes de comprar un paquete?\n2. ¿Cuánto cuesta la membresía mensual?\n3. ¿Tienen tapetes o llevo el mío?\n4. ¿Hay estacionamiento cerca del estudio?\n5. ¿Puedo congelar mi membresía si viajo?\n6. ¿Qué diferencia hay entre flow y yin?\n7. ¿Tengo que reservar antes?\n8. ¿La primera clase es gratis?\n9. ¿Qué pasa si cancelo tarde?\n10. ¿Tienen yoga prenatal?\n11. ¿Puedo compartir mi paquete con mi pareja?\n12. ¿Con cuánto tiempo de anticipación debo llegar?.\nAgrúpalas en los temas más comunes, hasta 8. Para cada tema, escribe la pregunta como la formulan los clientes y una respuesta de 2 a 4 oraciones usando solo estos datos: Estudio: Sauce Yoga, Calle Olmo 14, Ciudad de México. Clase suelta $250 MXN. Paquete de 10 clases $2,100 MXN, válido 4 meses, solo para una persona. Membresía mensual ilimitada $1,490 MXN; se puede congelar una vez al año hasta 30 días. La primera clase para alumnos nuevos cuesta $100 MXN. Los tapetes se prestan gratis. Hay que reservar en la app; cada clase es de 18 personas. Cancela al menos 6 horas antes o se cobra la clase. Las clases flow son dinámicas y siguen la respiración; en yin se sostienen las posturas de 3 a 5 minutos. Llega 10 minutos antes a tu primera clase..\nSi mis datos no responden una pregunta, escribe [FALTA RESPUESTA] en vez de inventar una.",
   "answer": "## 1. Precios y paquetes\n**\"¿Cuánto cuesta la membresía mensual?\"** / **\"¿Puedo compartir mi paquete con mi pareja?\"**\n\nLa membresía mensual ilimitada cuesta $1,490 MXN. El paquete de 10 clases cuesta $2,100 MXN, es válido por 4 meses y es solo para una persona, así que no se puede compartir con tu pareja.\n\n## 2. Primera clase y prueba\n**\"¿Puedo probar una clase antes de comprar un paquete?\"** / **\"¿La primera clase es gratis?\"**\n\nLa primera clase no es gratis, pero los alumnos nuevos la pueden tomar por $100 MXN. Una clase suelta regular cuesta $250 MXN.\n\n## 3. Reservas y cancelaciones\n**\"¿Tengo que reservar antes?\"** / **\"¿Qué pasa si cancelo tarde?\"**\n\nSí, las clases se reservan en la app, y cada clase tiene un cupo de 18 personas. Si cancelas con menos de 6 horas de anticipación, se cobra la clase.\n\n## 4. Llegada y estacionamiento\n**\"¿Con cuánto tiempo de anticipación debo llegar?\"** / **\"¿Hay estacionamiento cerca del estudio?\"**\n\nSi es tu primera clase, llega 10 minutos antes. Para las demás clases, [FALTA RESPUESTA]. El estudio está en Calle Olmo 14, Ciudad de México, pero sobre estacionamiento cercano: [FALTA RESPUESTA].\n\n## 5. Tapetes\n**\"¿Tienen tapetes o llevo el mío?\"**\n\nSí, tienes tapetes disponibles: se prestan gratis, así que no necesitas llevar el tuyo.\n\n## 6. Congelar la membresía\n**\"¿Puedo congelar mi membresía si viajo?\"**\n\nSí, puedes congelar tu membresía mensual una vez al año, hasta por 30 días. Esto te permite pausarla durante un viaje.\n\n## 7. Tipos de clase\n**\"¿Qué diferencia hay entre flow y yin?\"**\n\nLas clases flow son dinámicas y siguen el ritmo de la respiración. En las clases yin, las posturas se sostienen de 3 a 5 minutos cada una, con un ritmo más pausado.\n\n## 8. Yoga prenatal\n**\"¿Tienen yoga prenatal?\"**\n\n[FALTA RESPUESTA]",
   "format": "yes",
   "length": "partial",
   "invented": "none",
   "missing": "yes",
   "result": "minor",
   "note": "Spanish: 8 topics from the facts only, [FALTA RESPUESTA] for parking and prenatal yoga (and once more for arrival times after the first class), but the mats answer is one sentence (asked for 2 to 4).",
   "seconds": 11.9,
   "cost_usd": 0.001165
  },
  {
   "case": "T16ptbr",
   "locale": "pt-br",
   "prompt_id": "notes-to-actions",
   "job": "meetings-projects",
   "round": "t1",
   "prompt_version": "translated",
   "prompt": "A partir das anotações da reunião abaixo, liste:\n1) as decisões tomadas\n2) as tarefas numa tabela: tarefa, responsável, prazo\n3) as perguntas em aberto\nUse só nomes e datas que aparecem nas anotações. Escreva \"sem responsável\" ou \"sem data\" onde nada foi definido.\n\nAnotações:\nReunião de relançamento do site, 27 de outubro\nPresentes: Camila, Rafael, Bruna\n- Combinamos lançar em 17 de novembro, não em 10, porque as fotos dos produtos atrasaram\n- Rafael vai pegar as fotos finais com o fotógrafo até 7 de novembro\n- Bruna reescreve a página de frete, até sexta\n- Alguém precisa testar o checkout no celular\n- Camila vai perguntar à empresa de pagamentos sobre o Pix\n- Decidimos tirar o blog do lançamento\n- Em aberto: redirecionamos as URLs antigas de /loja? Ninguém sabia",
   "answer": "## 1) Decisões\n\n- Lançamento em **17 de novembro**, e não em 10, porque as fotos dos produtos atrasaram.\n- **Blog fica fora do lançamento.**\n\n## 2) Tarefas\n\n| Tarefa | Responsável | Prazo |\n|---|---|---|\n| Pegar as fotos finais com o fotógrafo | Rafael | 7 de novembro |\n| Reescrever a página de frete | Bruna | até sexta |\n| Testar o checkout no celular | sem responsável | sem data |\n| Perguntar à empresa de pagamentos sobre o Pix | Camila | sem data |\n\n## 3) Perguntas em aberto\n\n- Vamos redirecionar as URLs antigas de /loja? Ninguém sabia responder.",
   "format": "yes",
   "length": "n.a.",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "Portuguese: decisions, a 4-row table and the open question; 'sem responsavel' and 'sem data' where the notes give none.",
   "seconds": 6.7,
   "cost_usd": 0.000556
  },
  {
   "case": "T08ptbr",
   "locale": "pt-br",
   "prompt_id": "faq-from-tickets",
   "job": "customer-support",
   "round": "t1",
   "prompt_version": "translated",
   "prompt": "Aqui estão perguntas recentes de clientes: 1. Posso fazer uma aula experimental antes de comprar um pacote?\n2. Quanto custa o plano mensal?\n3. Vocês têm tapetes ou eu levo o meu?\n4. Tem estacionamento perto do estúdio?\n5. Posso congelar meu plano quando viajar?\n6. Qual a diferença entre flow e yin?\n7. Preciso reservar antes?\n8. A primeira aula é grátis?\n9. O que acontece se eu cancelar em cima da hora?\n10. Vocês têm yoga para gestantes?\n11. Posso dividir meu pacote com meu marido?\n12. Com quanto tempo de antecedência devo chegar?.\nAgrupe-as nos temas mais comuns, até 8. Para cada tema, escreva a pergunta do jeito que os clientes falam e uma resposta de 2 a 4 frases usando só estes fatos: Estúdio: Salgueiro Yoga, Rua das Olmeiras 14, São Paulo. Aula avulsa R$ 60. Pacote de 10 aulas R$ 500, válido por 4 meses, só para uma pessoa. Plano mensal ilimitado R$ 320; pode ser congelado uma vez por ano por até 30 dias. A primeira aula para alunos novos custa R$ 25. Os tapetes são emprestados de graça. É preciso reservar pelo app; cada aula tem 18 vagas. Cancele com pelo menos 6 horas de antecedência ou a aula é cobrada. As aulas de flow são dinâmicas e seguem a respiração; no yin as posturas ficam de 3 a 5 minutos. Chegue 10 minutos antes na primeira aula..\nSe os meus fatos não respondem a uma pergunta, escreva [FALTA RESPOSTA] em vez de inventar uma.",
   "answer": "**1. Aula experimental e primeira aula**\n*\"Dá pra fazer uma aula teste antes de comprar um pacote? E a primeira aula é de graça?\"*\nA primeira aula para alunos novos custa R$ 25, então não é gratuita. Esse valor reduzido funciona como a aula de teste antes de você decidir pelo pacote ou pelo plano.\n\n**2. Preço do plano mensal**\n*\"Quanto custa o plano mensal?\"*\nO plano mensal ilimitado custa R$ 320. Se preferir aulas avulsas, cada uma sai por R$ 60, ou você pode pegar o pacote de 10 aulas por R$ 500, válido por 4 meses e só para uma pessoa.\n\n**3. Congelar o plano em viagens**\n*\"Posso congelar meu plano quando eu viajar?\"*\nSim, o plano mensal pode ser congelado uma vez por ano, por até 30 dias. Isso serve para quem vai viajar ou ficar afastado por um período.\n\n**4. Diferença entre flow e yin**\n*\"Qual a diferença entre a aula de flow e a de yin?\"*\nAs aulas de flow são dinâmicas e seguem a respiração. No yin, as posturas ficam de 3 a 5 minutos cada, com um ritmo bem mais lento e contemplativo.\n\n**5. Reserva e cancelamento**\n*\"Preciso reservar antes? E o que acontece se eu cancelar em cima da hora?\"*\nSim, a reserva é feita pelo app, e cada aula tem 18 vagas. Para cancelar sem cobrança, avise com pelo menos 6 horas de antecedência; cancelamentos depois disso são cobrados como aula.\n\n**6. Pacote compartilhado**\n*\"Posso dividir meu pacote com meu marido?\"*\nNão. O pacote de 10 aulas é só para uma pessoa, então cada aluno precisa do seu próprio pacote. Se vocês quiserem treinar juntos, cada um deve ter o seu.\n\n**7. Chegada ao estúdio: tapete, estacionamento e horário**\n*\"Vocês têm tapetes ou eu levo o meu? Tem estacionamento perto? Com quanto tempo devo chegar?\"*\nOs tapetes são emprestados de graça, então não precisa levar o seu. Na primeira aula, chegue 10 minutos antes. Sobre estacionamento: [FALTA RESPOSTA]\n\n**8. Yoga para gestantes**\n*\"Vocês têm yoga para gestantes?\"*\n[FALTA RESPOSTA]",
   "format": "yes",
   "length": "yes",
   "invented": "none",
   "missing": "yes",
   "result": "pass",
   "note": "Portuguese: 8 topics, answers of 2 to 3 sentences from the facts only, [FALTA RESPOSTA] for parking and prenatal yoga.",
   "seconds": 12.3,
   "cost_usd": 0.001242
  },
  {
   "case": "T16fr",
   "locale": "fr",
   "prompt_id": "notes-to-actions",
   "job": "meetings-projects",
   "round": "t1",
   "prompt_version": "translated",
   "prompt": "À partir des notes de réunion ci-dessous, liste :\n1) les décisions prises\n2) les actions sous forme de tableau : tâche, responsable, échéance\n3) les questions ouvertes\nN'utilise que des noms et des dates qui figurent dans les notes. Écris « pas de responsable » ou « pas de date » quand aucun n'a été donné.\n\nNotes :\nPoint relance du site, 27 octobre\nPrésents : Claire, Julien, Inès\n- On lance le 17 novembre et non le 10, car les photos produits sont en retard\n- Julien récupère les photos finales auprès du photographe d'ici le 7 novembre\n- Inès réécrit la page livraison, pour vendredi\n- Quelqu'un doit tester le paiement sur mobile\n- Claire demande au prestataire de paiement pour Apple Pay\n- On supprime le blog pour le lancement\n- En suspens : faut-il rediriger les anciennes URL /boutique ? Personne ne savait",
   "answer": "# 1) Décisions prises\n\n- Le lancement du site est repoussé au **17 novembre** (et non le 10), car les photos produits sont en retard.\n- Le blog est **supprimé** pour le lancement.\n\n# 2) Actions\n\n| Tâche | Responsable | Échéance |\n|---|---|---|\n| Récupérer les photos finales auprès du photographe | Julien | 7 novembre |\n| Réécrire la page livraison | Inès | vendredi (pas de date précise) |\n| Tester le paiement sur mobile | pas de responsable | pas de date |\n| Demander au prestataire de paiement l'activation d'Apple Pay | Claire | pas de date |\n\n# 3) Questions ouvertes\n\n- Faut-il rediriger les anciennes URL /boutique ? Personne ne savait.",
   "format": "yes",
   "length": "n.a.",
   "invented": "minor",
   "missing": "yes",
   "result": "minor",
   "note": "French: 'pas de responsable' and 'pas de date' where the notes give none, but 'ask about Apple Pay' became 'ask for Apple Pay to be switched on'.",
   "seconds": 4.7,
   "cost_usd": 0.000415
  },
  {
   "case": "T08fr",
   "locale": "fr",
   "prompt_id": "faq-from-tickets",
   "job": "customer-support",
   "round": "t1",
   "prompt_version": "translated",
   "prompt": "Voici des questions récentes de clients : 1. Puis-je essayer un cours avant d'acheter une carte ?\n2. Combien coûte l'abonnement mensuel ?\n3. Vous prêtez des tapis ou j'apporte le mien ?\n4. Y a-t-il un parking près du studio ?\n5. Puis-je suspendre mon abonnement pendant un voyage ?\n6. Quelle est la différence entre flow et yin ?\n7. Faut-il réserver à l'avance ?\n8. Le premier cours est-il gratuit ?\n9. Que se passe-t-il si j'annule tard ?\n10. Proposez-vous du yoga prénatal ?\n11. Puis-je partager ma carte avec mon conjoint ?\n12. Combien de temps avant le cours dois-je arriver ?.\nRegroupe-les par sujets les plus fréquents, 8 au maximum. Pour chaque sujet, écris la question telle que les clients la formulent et une réponse de 2 à 4 phrases qui n'utilise que ces faits : Studio : Saule Yoga, 14 rue des Ormes, Lyon. Cours à l'unité 20 €. Carte de 10 cours 170 €, valable 4 mois, pour une seule personne. Abonnement mensuel illimité 99 €, suspendable une fois par an pendant 30 jours au plus. Le premier cours pour les nouveaux élèves coûte 10 €. Les tapis sont prêtés gratuitement. Réservation obligatoire dans l'appli ; 18 places par cours. Annulez au moins 6 heures avant, sinon le cours est décompté. Les cours de flow sont dynamiques et suivent le souffle ; en yin, on tient les postures 3 à 5 minutes. Arrivez 10 minutes en avance pour votre premier cours..\nSi mes faits ne répondent pas à une question, écris [RÉPONSE À FOURNIR] au lieu d'en inventer une.",
   "answer": "# Questions clients fréquentes, regroupées\n\n## 1. Tarifs et formules d'achat\n**Question client :** « Combien coûte l'abonnement mensuel ? »\n\nLe cours à l'unité coûte 20 €, la carte de 10 cours 170 € (valable 4 mois, pour une seule personne) et l'abonnement mensuel illimité 99 €. Le premier cours pour les nouveaux élèves coûte 10 €.\n\n## 2. Essai et premier cours\n**Question client :** « Puis-je essayer un cours avant d'acheter une carte ? »\n\nOui : le premier cours pour les nouveaux élèves coûte 10 €. Arrivez 10 minutes en avance pour ce premier cours. Pour réserver, passez par l'appli, car la réservation est obligatoire.\n\n## 3. Réservation\n**Question client :** « Faut-il réserver à l'avance ? »\n\nOui, la réservation est obligatoire et se fait dans l'appli. Chaque cours compte 18 places.\n\n## 4. Annulation\n**Question client :** « Que se passe-t-il si j'annule tard ? »\n\nIl faut annuler au moins 6 heures avant le cours. Passé ce délai, le cours est décompté.\n\n## 5. Suspension de l'abonnement\n**Question client :** « Puis-je suspendre mon abonnement pendant un voyage ? »\n\nOui, l'abonnement mensuel est suspendable une fois par an, pendant 30 jours au plus.\n\n## 6. Matériel et accès au studio\n**Question client :** « Vous prêtez des tapis ou j'apporte le mien ? »\n\nLes tapis sont prêtés gratuitement, vous n'avez donc pas besoin d'apporter le vôtre. Le studio se trouve au 14 rue des Ormes, à Lyon.\n\n**Question client :** « Y a-t-il un parking près du studio ? »\n\n[RÉPONSE À FOURNIR]\n\n## 7. Types de cours\n**Question client :** « Quelle est la différence entre flow et yin ? »\n\nLes cours de flow sont dynamiques et suivent le souffle. En yin, on tient les postures 3 à 5 minutes.\n\n## 8. Cours spécifiques\n**Question client :** « Proposez-vous du yoga prénatal ? »\n\n[RÉPONSE À FOURNIR]\n\n---\n\n**Non regroupées (les questions restantes) :**\n- **Carte partageable avec un conjoint** : « Puis-je partager ma carte avec mon conjoint ? » → La carte de 10 cours est pour une seule personne, donc non partageable. Je l'ai intégrée implicitement dans la section 1 mais elle mérite une réponse explicite. (Non traitée séparément faute de place dans les 8 sujets.)\n- **Arrivée avant le cours** : « Combien de temps avant le cours dois-je arriver ? » → Arrivez 10 minutes en avance pour votre premier cours. (Mentionnée dans la section 2.)",
   "format": "partial",
   "length": "partial",
   "invented": "none",
   "missing": "yes",
   "result": "minor",
   "note": "French: facts only and [REPONSE A FOURNIR] for parking and prenatal yoga, but it added a 'not grouped' list after the 8 topics and the freeze answer is one sentence.",
   "seconds": 8.8,
   "cost_usd": 0.000859
  }
 ],
 "dash_note": "Our site never prints the long dashes U+2013 and U+2014, so in this public copy each one in an answer is shown as a plain hyphen. The raw files in our repository keep the answers exactly as returned."
}
