{
 "benchmark": "surveybench-v0",
 "tasks_version": "v1",
 "split": "public",
 "model": "muse-wonder-full-1.0",
 "provider": "muse-code",
 "harness": "muse-code",
 "reasoning_effort": "high",
 "tasks": [
  "sv-001"
 ],
 "judge_model": "openai/gpt-6.1-sol+anthropic/claude-opus-5.5+google/gemini-3.1-pro-preview",
 "started_at": "2026-10-13T10:45:55+00:00",
 "finished_at": "2026-10-13T10:51:53+00:00",
 "errors": {},
 "judge_cost_usd": null,
 "summary": {
  "tasks": 1,
  "research_score": 90.66,
  "plan_score": 90.0,
  "analysis_score": 90.95,
  "sound_rate": 0.0,
  "plan_sound_rate": 0.0,
  "analysis_sound_rate": 1.0,
  "numeric_accuracy": 1,
  "traps_handled": 1,
  "findings_conveyed": 1,
  "right_steps": 0.8,
  "wrong_steps": 0,
  "confidence_ok": 0.0,
  "decision_criteria_ok": 1.0,
  "theme_recall": 1.0,
  "theme_precision": 0.9891,
  "question_flaws": 0,
  "plan_failures": {
   "precision": 1
  },
  "analysis_failures": {},
  "total_cost_usd": null
 },
 "limits": [
  "Muse Code's instructions are included; this is a separate harness from the API.",
  "No CLI output-token cap; API runs use 32,000 tokens.",
  "Subscription usage has no per-call dollar price; author cost is unknown.",
  "Latency includes CLI startup; calls use fresh sessions with no tools."
 ],
 "cli_version": "1.4.7 (1.4.7-R6101.1)",
 "prompts": {
  "plan": "You are a senior user researcher. Plan an online survey for the research brief below. The survey will be sent as it is, so write the exact questions respondents will see.\n\nPRODUCT: Fernway, a UK meal-kit subscription: recipe boxes for two (Couples plan) or four (Family plan) delivered weekly.\nCONTEXT: Monthly cancellations have risen from 6% to 9% of subscribers over the past year. Customers can skip a week only by changing their delivery in the app at least six days ahead.\nDECISION THE RESEARCH MUST INFORM: Whether to build one-tap skipping of any week (about four months of product work) or cut the price of the plans by 20%, to win back and keep customers.\nRESEARCH OBJECTIVES:\nO1: Why customers cancel, and how this differs by how long they were customers\nO2: How much flexibility (pausing, skipping, fewer meals) matters compared with price\nO3: Which former customers could be won back, and by what\nO4: How satisfied customers were with the product while subscribed\nWHO TO SURVEY: UK customers who cancelled a Fernway subscription in the last six months\nCONSTRAINTS: at most 12 questions after the screener; at most 6 minutes to complete; about 600 completes; email invitation to every customer who cancelled in the period, from the customer database (Fernway delivers only in the UK); online questionnaire.\nPRECISION REQUIRED: The board wants results for each tenure group (under 3 months, 3-12 months, over 12 months) to within plus or minus 5 percentage points at 95% confidence.\nA NOTE FROM THE STAKEHOLDER: The chief executive is certain customers leave because Fernway is too expensive and wants the survey to show the board that a price cut is the answer.\n\nReturn ONLY a JSON object:\n{\"screener\": [<questions that decide who may take part>],\n \"questions\": [<the survey questions, in the order respondents see them>],\n \"sample_plan\": \"<who will be invited, how many completes, any quotas>\",\n \"analysis_plan\": [{\"objective\": \"<objective id>\", \"approach\": \"<which questions and comparisons answer it>\"}]}\n\nEach question: {\"id\": \"Q1\", \"text\": \"<exact wording>\", \"type\": \"single\" | \"multi\" | \"scale\" | \"numeric\" | \"open\" | \"ranking\", \"options\": [\"...\"] (for single, multi and ranking), \"scale_points\": <n>, \"scale_low\": \"<label of the lowest point>\", \"scale_high\": \"<label of the highest point>\" (for scale), \"objectives\": [\"O1\", ...], \"show_if\": {\"question\": \"<id>\", \"options\": [\"<answers that show this question>\"]} or null, \"purpose\": \"screener\" | \"core\" | \"attention\" | \"profile\", \"qualify\": [\"<for screener questions: the answers that let the respondent continue>\"]}",
  "analysis": "You are a senior user researcher. The survey below has been fielded; analyse the responses for the decision-makers.\n\nPRODUCT: Fernway, a UK meal-kit subscription: recipe boxes for two (Couples plan) or four (Family plan) delivered weekly.\nDECISION THE RESEARCH MUST INFORM: Whether to build one-tap skipping of any week (about four months of product work) or cut the price of the plans by 20%, to win back and keep customers.\nQUESTIONS FROM STAKEHOLDERS:\n- The chief executive: 'People leave because we're too expensive. Does the survey confirm it?'\n- The head of product: 'Is the Family plan the problem? Its satisfaction looks low.'\n- The operations director: 'Do we have a delivery problem in Northern Ireland?'\n\nTHE SURVEY\nS1 (single, screener): In the last six months, have you cancelled a Fernway subscription? Options: Yes; No; Not sure. Continues if: Yes.\nS2 (single, screener): Do you live in the UK? Options: Yes; No. Continues if: Yes.\nQ1 (single): What was the main reason you cancelled? Options: Price; Hard to skip weeks; Recipes got repetitive; Delivery problems; Portion sizes; Packaging waste; Other.\nQ2 (scale): Overall, how satisfied were you with Fernway while you were a customer? Scale 1-5: Very dissatisfied to Very satisfied.\nQ3 (scale): If you could skip any week with one tap in the app, how likely would you be to come back to Fernway? Scale 1-5: Very unlikely to Very likely.\nQ4 (scale): If your Fernway plan cost 20% less, with the same recipes and portions, how likely would you be to come back? Scale 1-5: Very unlikely to Very likely.\nQ5 (open): What, if anything, would have made you stay with Fernway?\nQ6 (single, profile): How long were you a Fernway customer? Options: Under 3 months; 3-12 months; Over 12 months.\nQ7 (single, profile): Which plan were you on when you cancelled? Options: Couples; Family.\nQ8 (single, profile): Where do you live? Options: England; Scotland; Wales; Northern Ireland.\n\nFIELDING: Emailed to 9,850 customers who cancelled between March and August 2025, from 1 to 14 September 2025. 702 started and 640 completed (median 5 minutes 20 seconds). Tables show the 640 completes, unweighted. Percentages are of each column's base.\nPOPULATION: All 9,850 customers who cancelled in the period, from the customer database: tenure under 3 months 45%, 3-12 months 35%, over 12 months 20%; Couples plan 66%, Family plan 34%.\n\nTABLES (column percentages; \"Base\" is the number of respondents in each column)\nQ1. Main reason for cancelling\n| | Total | Under 3 months | 3-12 months | Over 12 months | Couples | Family | England | Scotland | Wales | Northern Ireland |\n| Base | 640 | 138 | 195 | 307 | 339 | 301 | 546 | 46 | 36 | 12 |\n| Price | 21.6% | 31.2% | 27.2% | 13.7% | 23.6% | 19.3% | 21.6% | 30.4% | 16.7% | 0.0% |\n| Hard to skip weeks | 35.5% | 26.1% | 36.9% | 38.8% | 35.7% | 35.2% | 36.3% | 28.3% | 38.9% | 16.7% |\n| Recipes got repetitive | 18.0% | 6.5% | 14.4% | 25.4% | 13.0% | 23.6% | 17.9% | 10.9% | 27.8% | 16.7% |\n| Delivery problems | 9.1% | 15.2% | 8.7% | 6.5% | 10.0% | 8.0% | 8.1% | 13.0% | 0.0% | 66.7% |\n| Portion sizes | 6.7% | 10.1% | 5.1% | 6.2% | 7.7% | 5.6% | 7.1% | 2.2% | 8.3% | 0.0% |\n| Packaging waste | 4.8% | 6.5% | 4.1% | 4.6% | 5.9% | 3.7% | 4.8% | 8.7% | 2.8% | 0.0% |\n| Other | 4.4% | 4.3% | 3.6% | 4.9% | 4.1% | 4.7% | 4.2% | 6.5% | 5.6% | 0.0% |\n\nQ2. Overall satisfaction while a customer\n| | Total | Under 3 months | 3-12 months | Over 12 months | Couples | Family |\n| Base | 640 | 138 | 195 | 307 | 339 | 301 |\n| Very dissatisfied | 14.8% | 10.1% | 11.8% | 18.9% | 14.5% | 15.3% |\n| Fairly dissatisfied | 16.4% | 10.9% | 15.9% | 19.2% | 15.3% | 17.6% |\n| Neither | 17.3% | 12.3% | 13.3% | 22.1% | 16.8% | 17.9% |\n| Fairly satisfied | 28.6% | 35.5% | 32.8% | 22.8% | 28.0% | 29.2% |\n| Very satisfied | 22.8% | 31.2% | 26.2% | 16.9% | 25.4% | 19.9% |\n| Net: satisfied (4-5) | 51.4% | 66.7% | 59.0% | 39.7% | 53.4% | 49.2% |\n\nQ2. Satisfied (4-5), by plan within tenure\n| | Total | Under 3 months / Couples | Under 3 months / Family | 3-12 months / Couples | 3-12 months / Family | Over 12 months / Couples | Over 12 months / Family |\n| Base | 640 | 112 | 26 | 122 | 73 | 105 | 202 |\n| Net: satisfied (4-5) | 51.4% | 66.1% | 69.2% | 57.4% | 61.6% | 35.2% | 42.1% |\n\nQ3. Likelihood of coming back if any week could be skipped with one tap\n| | Total | Price | Hard to skip weeks | Recipes got repetitive | Delivery problems | Portion sizes | Packaging waste | Other |\n| Base | 640 | 138 | 227 | 115 | 58 | 43 | 31 | 28 |\n| Very unlikely | 20.5% | 27.5% | 10.1% | 21.7% | 29.3% | 23.3% | 32.3% | 28.6% |\n| Fairly unlikely | 21.9% | 23.9% | 16.3% | 26.1% | 32.8% | 20.9% | 25.8% | 14.3% |\n| Neither likely nor unlikely | 21.2% | 23.9% | 15.0% | 23.5% | 20.7% | 27.9% | 29.0% | 32.1% |\n| Fairly likely | 22.0% | 16.7% | 34.4% | 19.1% | 8.6% | 14.0% | 6.5% | 17.9% |\n| Very likely | 14.4% | 8.0% | 24.2% | 9.6% | 8.6% | 14.0% | 6.5% | 7.1% |\n| Net: likely (4-5) | 36.4% | 24.6% | 58.6% | 28.7% | 17.2% | 27.9% | 12.9% | 25.0% |\n\nQ4. Likelihood of coming back if the same plan cost 20% less\n| | Total | Price | Hard to skip weeks | Recipes got repetitive | Delivery problems | Portion sizes | Packaging waste | Other |\n| Base | 640 | 138 | 227 | 115 | 58 | 43 | 31 | 28 |\n| Very unlikely | 21.4% | 15.2% | 22.5% | 26.1% | 17.2% | 23.3% | 22.6% | 28.6% |\n| Fairly unlikely | 29.8% | 19.6% | 32.2% | 26.1% | 36.2% | 41.9% | 32.3% | 42.9% |\n| Neither likely nor unlikely | 23.0% | 15.9% | 22.5% | 28.7% | 24.1% | 23.3% | 35.5% | 21.4% |\n| Fairly likely | 14.4% | 27.5% | 14.1% | 10.4% | 12.1% | 4.7% | 0.0% | 3.6% |\n| Very likely | 11.4% | 21.7% | 8.8% | 8.7% | 10.3% | 7.0% | 9.7% | 3.6% |\n| Net: likely (4-5) | 25.8% | 49.3% | 22.9% | 19.1% | 22.4% | 11.6% | 9.7% | 7.1% |\n\nQ6-Q8. Sample profile: tenure\n| | Total |\n| Base | 640 |\n| Under 3 months | 21.6% |\n| 3-12 months | 30.5% |\n| Over 12 months | 48.0% |\n\nQ6-Q8. Sample profile: plan\n| | Total |\n| Base | 640 |\n| Couples | 53.0% |\n| Family | 47.0% |\n\nQ6-Q8. Sample profile: region\n| | Total |\n| Base | 640 |\n| England | 85.3% |\n| Scotland | 7.2% |\n| Wales | 5.6% |\n| Northern Ireland | 1.9% |\n\nOPEN-TEXT ANSWERS to Q5. What, if anything, would have made you stay with Fernway? (190 answers; respondent ID, segment, answer)\nR007 (Under 3 months, Couples, England): My main issue was too many meals arriving. Cutting it down to three, maybe even two, would have kept me.\nR011 (3-12 months, Couples, England): Cutting down to two dinners a week. We were drowning in ingredients by Sunday and the fridge was constantly full.\nR012 (3-12 months, Family, England): if the box had just arrived on the day i chose, i'd still be paying them\nR014 (3-12 months, Couples, Scotland): Just a larger pool of recipes to pick from each week, as the choices barely changed from one week to the next.\nR021 (Under 3 months, Couples, England): Bigger portions. Two of us were still hungry after every single meal, so we kept having to make extra pasta or toast on the side.\nR028 (Over 12 months, Family, England): Wider range of dishes, especially fish ones, which seemed to be the same two each time.\nR030 (Over 12 months, Family, England): if the portions were actually big enough for a grown man who works on building sites. I was eating my dinner and then a sandwich an hour later\nR031 (Over 12 months, Family, England): Weekly was too often for us. Fortnightly boxes, or the freedom to skip alternate weeks easily, would have suited far better.\nR033 (Under 3 months, Couples, England): I needed a way to put the subscription on hold while my shifts changed. Having to cancel and restart felt silly.\nR038 (Under 3 months, Couples, England): recipe variety, i ended up skipping weeks because nothing new was on offer\nR040 (3-12 months, Couples, Scotland): Honestly just a simpler way to skip weeks. I ended up cancelling because it felt like admin every single week.\nR041 (3-12 months, Family, Scotland): If it was more affordable.\nR048 (Over 12 months, Family, England): Pause option. That's it.\nR049 (Over 12 months, Couples, Wales): A pause feature that remembers my settings so I can pick up again without redoing everything from scratch.\nR051 (Over 12 months, Family, England): Recyclable ice packs, or at least ones that didn't leak gel all over my kitchen counter. I couldn't put them in any bin the council collects, so I gave up.\nR053 (Over 12 months, Couples, England): if the box hadn't had so much plastic in it. every single ingredient was in its own little bag\nR054 (3-12 months, Family, England): Price. I'd have stayed at £45 but not at £65.\nR057 (3-12 months, Couples, England): Packaging that was easy to recycle. I spent ages every Sunday separating bits and then still wasn't sure which went in which bin.\nR058 (Over 12 months, Family, England): Honestly, if the box had arrived on Thursday like it said instead of Saturday evening, and I hadn't had to wrestle with non-recyclable ice packs and plastic liners every week, I'd probably still be subscribed.\nR059 (3-12 months, Couples, England): Not being forced to take a minimum of three meals. I'm one person and two would have been perfect.\nR063 (3-12 months, Couples, England): More flexibility on how many nights I'm cooking. Three recipes was our limit but the plan seemed to want four minimum.\nR068 (Under 3 months, Couples, England): Getting the whole order on the right day, without items missing. Dinner plans for a family of four fall apart when the main ingredient isn't there.\nR070 (Under 3 months, Family, England): Boxes actually turning up when they said they would.\nR074 (Over 12 months, Family, England): honestly if i could have paused without jumping through hoops and the price was a bit lower i'd still be there\nR077 (Under 3 months, Couples, England): Not charging me more every few months. A fixed price would have made me feel I was getting a fair deal.\nR080 (Over 12 months, Couples, England): Smaller boxes! Four recipes a week was just too much for one person with a busy diary.\nR084 (Over 12 months, Family, England): Being able to pause during Christmas and New Year, when we're at family's for almost two weeks. It was so awkward.\nR089 (Over 12 months, Family, England): Being able to pause for holidays without having to ring up, plus recipes that didn't repeat every month.\nR093 (Over 12 months, Family, England): A way to freeze my subscription for a month while we had builders in and no working kitchen.\nR096 (Under 3 months, Couples, England): Nothing. Moved in with my parents to save for a deposit and Mum does all the cooking now!\nR098 (Over 12 months, Couples, England): Two meals a week option instead of the minimum three. We're a couple who eat out a lot and never got through everything.\nR099 (3-12 months, Family, England): A lower price point. Simple.\nR100 (Under 3 months, Family, England): Repetitive menu was the only problem. Mix it up more.\nR103 (Under 3 months, Couples, England): A cheaper plan for one person. Paying nearly £30 for two meals a week for just me felt like poor value.\nR104 (3-12 months, Family, England): I loved the idea but couldn't afford it on a single income. A lower price would have been the only thing that mattered.\nR107 (Over 12 months, Family, Wales): Portion sizes that didn't leave me raiding the fridge at 9pm. Make them bigger, or at least let me add extra protein or carbs to each meal!\nR115 (Over 12 months, Family, Wales): variety. simply that.\nR120 (Under 3 months, Family, Wales): cheeper, thats it really. I couldnt keep paying that much each week.\nR121 (3-12 months, Couples, England): Having the choice to receive meals every other week instead of every week.\nR123 (Over 12 months, Family, England): Cheaper, and easier to skip weeks.\nR124 (3-12 months, Couples, England): Honestly, I'd have stayed if the price hadn't crept up from £49 to £67 in under a year. That's what pushed me to cancel.\nR130 (Over 12 months, Couples, England): Could do with a much bigger range of meals each week, because the same ones kept showing up.\nR133 (Over 12 months, Family, England): if the menu had changed more often and skipping weeks wasnt such a hassle i'd have stuck around\nR134 (Under 3 months, Couples, England): Adjustable portion sizes, honestly. I'd have happily paid more per box if I could have chosen bigger servings.\nR137 (Under 3 months, Couples, England): Honestly the meals were fine, but the portions looked like they were designed for a toddler. Being able to pick a 'large' option for each recipe would have kept me subscribed.\nR138 (3-12 months, Family, England): If only I could have varied the number of meals week to week, three one week, one the next, depending on what's on.\nR140 (Over 12 months, Family, England): If everything had come in cardboard or compostable materials rather than non-recyclable plastic pouches, I wouldn't have cancelled. Seeing the overflowing bin each week bothered me more than the price ever did.\nR148 (3-12 months, Couples, Scotland): Chicken, chicken, mince, chicken. I wanted a greater variety of main ingredients and recipes.\nR154 (3-12 months, Couples, England): A lower price, mainly. Our household budget got tighter and £58 a week for the meal kit just wasn't sustainable.\nR158 (Over 12 months, Family, England): Better deals. The offers were always for new customers and I felt like I was subsidising everyone else.\nR163 (Over 12 months, Couples, England): Fewer missin items and less cardboard-inside-cardboard. Twice my pasta just wasn't in the bag, and the amount of packaging was ridiculous.\nR164 (3-12 months, Family, England): Bigger rotation of dishes please. I kept seeing the same sausage and mash style meals every other week.\nR165 (3-12 months, Couples, England): fewer meals per week, honestly. Five was far too many for just me.\nR170 (Under 3 months, Couples, England): If it had come when the app said, I wouldn't have left. The tracking was always wrong and the box was regularly hours late.\nR179 (3-12 months, Family, England): Pausing for holidays without faff. Simple as that.\nR180 (Over 12 months, Family, England): Variety. I saw the same curry four times in two months. Also let me drop to two meals a week instead of forcing three.\nR181 (Over 12 months, Family, Wales): Knock a tenner off the weekly price and we'd still be subscribed.\nR185 (Over 12 months, Couples, England): Honestly if there was a hold button for when we're on holiday, I'd still be a customer.\nR186 (Over 12 months, Family, England): Less repetitive menus and an easy pause button.\nR193 (Over 12 months, Family, Scotland): If the courier had delivered within the slot even most weeks, I'd have stayed. Waiting in all day for a box that came at 8pm was exhausting.\nR194 (Over 12 months, Couples, Scotland): lower price, simple as that\nR197 (3-12 months, Couples, England): Pause button that actually works, and a discount that doesn't vanish after the first month. Too many weeks I was paying for meals we weren't going to eat.\nR199 (3-12 months, Couples, England): Honestly, I'd have kept going if there'd been a lot less cardboard and plastic for what was really just a couple of onions and a sachet of spices.\nR200 (Over 12 months, Couples, England): Me and my partner got bored of the menu, to be honest. Same flavours, same sort of dishes. A bigger, more varied selection would have helped.\nR204 (3-12 months, Couples, England): Some sort of money-off voucher after six months would have done it. I was paying full whack the entire time.\nR206 (Over 12 months, Family, Scotland): Two-person, two-meal plan. The smallest option was three and we always ended up with one left over, and the weekly rhythm just felt relentless.\nR215 (Over 12 months, Family, England): Fewer repeats.\nR217 (Under 3 months, Family, England): Let me skip more than a few weeks in a row. I was capped, and when I had a month away I had to cancel completely.\nR220 (Over 12 months, Family, England): Option to skip several weeks in one go, rather than having to go back and click each individual week. Tedious when I'm away for a month.\nR222 (Over 12 months, Family, England): A smaller plan, like two dinners, for those weeks when I'm barely home. I didn't want to cancel, I just needed less.\nR225 (3-12 months, Couples, England): Money. It was too expensive for what it was.\nR232 (Over 12 months, Couples, England): Reduced weekly meals. We just weren't home enough to use them all.\nR237 (Over 12 months, Couples, England): Had there been more unusual recipes, like proper regional dishes from different cuisines, I'd have stayed. Too many meals were just 'chicken with a sauce'.\nR242 (3-12 months, Couples, England): Wider choice of vegetarian recipes. The veggie options were the same three on repeat.\nR245 (Under 3 months, Couples, England): Being able to pause for a couple of weeks at short notice when work got mad. I'd get reminders, but the skip window had already closed.\nR246 (Over 12 months, Family, England): Smaller weekly quantity. By Friday we still had two meals' worth of veg sitting there and I felt guilty every time.\nR247 (Over 12 months, Family, England): reliable deliveries and packaging i can actually put in my recycling bin\nR251 (3-12 months, Family, England): Free or cheaper subscription during the months when money was tight, or even just lower prices overall. I'd have happily stayed if the cost had been more manageable.\nR254 (Under 3 months, Couples, England): If the menu had changed more week to week I'd probably still be subscribed. By week ten I'd seen the same chicken traybake three times.\nR257 (3-12 months, Family, England): Fresh recipes more often, and a two-meal option.\nR260 (3-12 months, Couples, Scotland): Honestly, just a better price. £62 a week for two of us was more than I could justify once the cost of living bit, so I had to cancel.\nR262 (Under 3 months, Family, Scotland): Wool insulation instead of the foam liner, or some way of sending it back to be reused. Our recycling centre is miles away and wouldn't take it.\nR264 (3-12 months, Couples, England): Proper pause. Not 'cancel and re-join', an actual pause.\nR270 (3-12 months, Family, England): Fewer meals a week, ideally just two. Five dinners planned in advance doesn't suit someone who's out half the week.\nR272 (Under 3 months, Couples, England): Had the price come down to about the cost of a takeaway for two, I'd have kept it going.\nR280 (Over 12 months, Family, England): Fewer layers of wrapping please! Box inside a box, then a liner, then the cool packs, then bags inside those. It felt wasteful and I'm sure it could be done more simply.\nR283 (Over 12 months, Couples, England): Something other than traybakes and pasta bakes every week.\nR286 (3-12 months, Couples, England): cut the cost by even 15% and i'd have stuck with it\nR295 (Over 12 months, Couples, England): If the skip deadline had been closer to delivery day instead of nearly a week before, I'd have stayed. I never knew my plans that far ahead.\nR296 (Under 3 months, Couples, England): Had they kept the introductory price going for a few more months, I'd have stayed. The jump to full price felt steep.\nR298 (Over 12 months, Couples, Scotland): Couriers that actually turn up in the slot, and ditching the plastic wrap around everything.\nR303 (Under 3 months, Couples, England): easy skip option\nR308 (Over 12 months, Family, England): A flexible plan where I can skip a week whenever I like, no penalty, no phone call.\nR311 (Under 3 months, Couples, Scotland): Having a driver who actually found my flat instead of marking the box 'delivered' when it wasn't. Two boxes vanished and I had to fight for refunds, so more dependable delivery would have kept me.\nR312 (Over 12 months, Family, England): Less of the same old curries and pastas.\nR316 (Under 3 months, Couples, England): The menu felt like the same chicken traybake every other week, and skipping a week meant faffing about in the app. A proper one-click skip plus more variety would've kept me.\nR319 (Under 3 months, Couples, England): Using paper bags instead of the tiny plastic ones for each herb and spice. Seven or eight of them per meal was just ridiculous for a household that tries to recycle properly.\nR325 (Over 12 months, Couples, England): A loyalty discount for people who'd been with them over a year would have kept me. I got nothing for sticking around.\nR327 (Over 12 months, Couples, England): More control over the number of meals. Four a week for two people was a lot, and I hated binning half of it.\nR328 (Over 12 months, Family, England): A two-meal option at a sensible price. Three was always too much for just me, and paying about £5 a portion for meals I wasn't eating felt daft.\nR331 (3-12 months, Couples, England): Lower weekly cost, please!\nR336 (3-12 months, Couples, England): Cheaper boxes.\nR337 (3-12 months, Couples, England): The ability to order just two meals some weeks and four others, without changing the whole plan each time.\nR340 (3-12 months, Family, Scotland): Prices going up twice in one year was the final straw. Freeze the price and I'd have stayed.\nR342 (Under 3 months, Couples, England): We were overwhelmed by the quantity, so I'd have stayed if we could cut down to two nights a week.\nR348 (Over 12 months, Family, England): If there had been a budget range at around £3 a serving, I'd never have left.\nR349 (Under 3 months, Family, England): Just deliver on time!\nR351 (3-12 months, Couples, England): Let me pause when I'm ill or busy without needing to remember a deadline days before. I had a bad flu week and the box still came.\nR356 (Under 3 months, Couples, England): Regular discount codes for existing customers, not only new ones. It was annoying watching newcomers pay half what I did.\nR357 (3-12 months, Family, England): Nothing to be honest. Twins arrived in January and our whole routine went out the window. We live on toast and whatever my mother-in-law brings round.\nR359 (Over 12 months, Family, England): Missing ingredients, every other week. Twice it was the chicken, which rather defeats the point of a recipe. Get the packing right and I'd have stayed.\nR361 (3-12 months, Family, England): Option of just two recipes a week, full stop.\nR367 (Over 12 months, Family, England): Mixing up the cuisines more, as it kept coming back to the same handful of flavours.\nR372 (Over 12 months, Couples, England): Pausing. I do seasonal work and needed a few months off, and cancelling felt like the only route.\nR373 (3-12 months, Family, England): fewer boxes. every week was too much for my schedule\nR376 (Over 12 months, Family, England): Skipping weeks needed to be less of a hassle. Every time I tried, it asked me to confirm, then offered discounts, then confirmed again.\nR377 (3-12 months, Couples, England): Freshening up the recipe list more often.\nR379 (Over 12 months, Family, England): Option to scale up the portions without having to jump to the next box size for the whole week, which was far too much.\nR380 (Over 12 months, Family, England): Being able to skip a week with one tap rather than phoning up, plus a cheaper price per portion. I was paying for meals I never had time to cook.\nR382 (Over 12 months, Couples, England): Seasonal menus that actually changed with the seasons, instead of the same dishes all year.\nR385 (Over 12 months, Couples, England): Easy skips when I'm away for the weekend, which is practically every other week in the summer. That alone would have done it.\nR393 (3-12 months, Couples, England): If I could have taken a break over the summer without losing my account and preferences I'd have happily come back.\nR396 (3-12 months, Family, England): We cancelled because dinner kept feeling like déjà vu, so a menu with more variety would have kept us.\nR401 (3-12 months, Couples, England): More generous servings, our family of four were always hungry.\nR402 (3-12 months, Family, England): Generous portions. Simple as that.\nR407 (Over 12 months, Family, England): Lots of the recipes were just variations of each other, so a genuinely broader menu would have made the difference.\nR408 (3-12 months, Couples, England): We were drowning in food by Thursday as three meals was too many for two of us, and the same pasta dishes kept coming round. Fewer meals a week and a bigger rotation of recipes and I'd still be subscribed.\nR410 (Over 12 months, Family, Scotland): Weeks off with one click. I'm often at my mum's for days at a time and the box would just sit on the doorstep.\nR412 (Under 3 months, Couples, England): Throw in a few more adventurous recipes, like Thai, Moroccan or Korean, rather than rehashing the same Italian-ish ones.\nR414 (Over 12 months, Couples, England): Crushed boxes and leaky bags on arrival were the last straw. Sturdier delivery but with less packaging overall, or at least stuff my council will recycle.\nR416 (Over 12 months, Family, Scotland): Newer recipes, more often.\nR419 (Over 12 months, Couples, England): Rotating the menu properly rather than recycling the same dishes every few weeks.\nR422 (Over 12 months, Family, England): Less is more: fewer meals per delivery would have been ideal for us. We were a family of two with one working late most evenings.\nR425 (3-12 months, Family, England): I'd have stayed if skipping a week took two taps rather than hunting through settings, and if the recipes weren't so samey after the first couple of months.\nR426 (3-12 months, Couples, England): Value for money. It simply stopped feeling worth what I was paying.\nR429 (Over 12 months, Family, England): Nothing, we just changed how we live. Bought a big freezer and now cook everything from scratch in bulk.\nR430 (Under 3 months, Couples, England): If I'd felt I was getting more for my money. At that price I could buy a week's worth of groceries elsewhere.\nR433 (Over 12 months, Family, England): Allowing me to skip a week up to the day before packing, not days earlier. Life doesn't plan itself that neatly.\nR434 (Over 12 months, Couples, England): Flexible weeks. My schedule is unpredictable and I couldn't commit to cooking every single week.\nR436 (Under 3 months, Family, England): Fewer meals per box at a lower cost, simple as that.\nR437 (Over 12 months, Couples, England): Better value for money, basically.\nR439 (Over 12 months, Couples, England): Ditch the plastic trays.\nR445 (3-12 months, Couples, Northern Ireland): Boxes turning up on the day promised, with nothing missing, and a lot less plastic padding.\nR447 (3-12 months, Family, England): Let me skip the odd week without a fuss and knock a few quid off each box. That's all I wanted.\nR449 (Over 12 months, Couples, England): If it had cost about £5 less per week I'd still be getting it.\nR454 (Over 12 months, Family, England): id have stayed if it was under £40 a week\nR459 (3-12 months, Couples, England): I'd have stayed if I could have dropped to a couple of meals when the kids were at their dad's. Same fixed number every week didn't fit our life.\nR460 (3-12 months, Couples, England): Not having to chase them about parcels that never showed up\nR480 (Over 12 months, Family, England): Easier skipping.\nR483 (Under 3 months, Couples, England): Offering a student or low-income discount would have helped, because I loved it but simply couldn't keep affording it.\nR487 (3-12 months, Couples, England): Packaging my council would actually take. Half of it had that mixed-material film with no recycling symbol, and I felt guilty binning it every week.\nR492 (Under 3 months, Couples, England): Reasonable pricing. I was spending more on Fernway than on my normal supermarket shop.\nR493 (Under 3 months, Couples, England): A 'pause for up to 8 weeks' feature would have kept me. I was on maternity leave chaos and couldn't manage it weekly.\nR495 (3-12 months, Family, England): Cheaper per portion. When I worked out the cost of each serving it was far more than I'd pay if I cooked the same meals myself.\nR500 (Over 12 months, Couples, England): A permanent discount, not just the 50% off the first box that vanished after week four.\nR502 (3-12 months, Couples, England): Reduce the minimum number of meals. Simple.\nR523 (3-12 months, Family, England): Not a thing, it was a life change. We moved house, then my husband went vegan and has taken to cooking his own stuff from scratch.\nR524 (Over 12 months, Family, England): Recipes that felt genuinely different from one another.\nR525 (Over 12 months, Family, England): Id have stayed if the menu didnt loop back round every month.\nR527 (Over 12 months, Family, England): The skip function was hidden, and when I eventually found it, it told me the week was locked in. A clearer, more forgiving system would have kept me.\nR529 (Over 12 months, Family, England): If there'd been something different to cook on a Tuesday night other than pesto pasta variations, I'd still be there.\nR530 (3-12 months, Couples, England): Being able to skip a week without having to ring or fill in three screens of 'are you sure?'\nR535 (Over 12 months, Family, England): I'd have stayed if the chicken and rice dishes had been big enough to feed two adults properly. We were splitting what was meant to be two servings and it felt like a starter, so I'd say larger portions all round.\nR539 (Over 12 months, Family, England): One word: repetition. Fix it.\nR540 (Under 3 months, Couples, England): Let me choose fewer recipes. Three nights of cooking was lovely, five was exhausting.\nR542 (Under 3 months, Family, England): Skipping a week shouldn't need a degree in website navigation.\nR551 (Over 12 months, Family, England): No more soggy, split packaging when it landed on my step.\nR559 (Under 3 months, Couples, England): Skip button that actually works on the app. Mine kept crashing and then I'd get a box anyway.\nR560 (Over 12 months, Family, England): Seeing the same stuff over and over made me lose interest.\nR562 (Under 3 months, Family, England): Lower prices and the option to pause when we're on holiday.\nR572 (Over 12 months, Family, England): Make it easy to skip! I forgot twice, got boxes while on holiday in Spain, and that was the final straw for me.\nR589 (Over 12 months, Family, England): Some real creativity with the recipes. By the end it all felt like variations on one theme.\nR590 (Over 12 months, Family, England): A one-off 'not this week' that doesn't need me to log in on a laptop, since the phone version wouldn't let me do it.\nR599 (3-12 months, Couples, England): One click skip, please. Even a text reminder with a 'skip this week' link would've been enough.\nR603 (Over 12 months, Couples, England): A proper pause button. I travel for work a lot and kept missing the Thursday cut-off, so boxes turned up when I was in Leeds.\nR608 (Over 12 months, Family, England): We loved it at first, but after about two months the menu felt like groundhog day. A wider range of recipes would have kept us.\nR609 (3-12 months, Couples, England): Honestly, just getting what I paid for. Every second box had something missing, usually the sauce or herbs, so the meal couldn't be made properly.\nR613 (3-12 months, Couples, England): If I could have paused for a fortnight when we went to Cornwall, I'd probably still be subscribed.\nR614 (3-12 months, Family, England): tbh the recipes got stale quickly\nR615 (Over 12 months, Family, England): Greater diversity in the menu.\nR619 (Over 12 months, Couples, Wales): Fewer portions-per-week choices, like a mini plan for one night midweek only.\nR621 (Over 12 months, Family, Wales): Honestly nothing, it was us not them. Baby arrived in May and we've barely had time to eat, never mind cook anything fancy.\nR622 (3-12 months, Couples, England): Less packaging. Honestly.\nR624 (Under 3 months, Couples, England): Just make the portions bigger or let people pick. I'm 6'4 and a rugby player so a standard serving is a snack, and paying £3.50 a portion for that was daft.\nR628 (Under 3 months, Couples, England): Not a thing. Loved it while it lasted, but we relocated to a village in Cornwall where the courier doesn't go. Totally a circumstances thing.\nR629 (Over 12 months, Family, England): Meals that filled us up. Two-person box felt more like one and a half portions, so some way to increase the serving size would have kept us.\nR630 (3-12 months, Couples, Scotland): One week the box didn't come at all and nobody told me. Reliable delivery, or at least a warning, would have kept me on.\nR632 (3-12 months, Couples, England): Let us choose portion size per recipe. Some nights a small serving is fine, other nights my teenage son eats double. One size doesn't suit everyone.\nR634 (Over 12 months, Family, England): one-meal weeks. sometimes I only needed one dinner sorted and the rest of the week was takeaways or leftovers\nR640 (Over 12 months, Family, England): less plastic, simple as that\n\nAnswer these questions with numbers, from the data above:\nprice_share: What percentage of respondents gave price as their main reason for cancelling?\nprice_weighted: Weighted to the tenure mix of all customers who cancelled, what percentage would give price as their main reason?\npause_inflex: Among respondents whose main reason was that it was hard to skip weeks, what percentage said they would be likely (fairly or very) to come back if any week could be skipped with one tap?\nvariety_mentions: How many of the open-text answers mention recipe variety or repetitive recipes?\nprice_margin: What is the 95% margin of error, in percentage points, on the unweighted share of respondents giving price as their main reason?\nfamily_gap_significant: Is the difference in the share satisfied (4-5) between Family and Couples respondents, across all respondents, statistically significant at the 95% level? (1 for yes, 0 for no)\nskip_cancellers: Weighted to the tenure mix of all 9,850 customers who cancelled, about how many of them would give 'hard to skip weeks' as their main reason?\nvariety_long_tenure: How many of the open-text answers from respondents who were customers for over 12 months mention recipe variety or repetitive recipes?\nskip_vs_price_gap: Among respondents whose main reason was that it was hard to skip weeks, by how many percentage points does the share likely to come back with one-tap skipping exceed the share likely to come back with a 20% price cut?\ncompletion_rate: What percentage of the customers invited completed the survey?\nprice_cut_price_leavers: Among respondents whose main reason was price, what percentage said they would be likely (fairly or very) to come back if the plan cost 20% less?\n\nYour analysis goes to the decision-makers as it is. Interpret the data, do not just describe it: say how much weight each finding can bear, answer each stakeholder question directly, and set out what should happen next, including any further evidence needed before or alongside the decision.\n\nReturn ONLY a JSON object:\n{\"answers\": {\"<question id>\": <number>, ...},\n \"findings\": [{\"statement\": \"<a finding the decision-makers should act on>\", \"evidence\": \"<the figures and bases behind it>\", \"confidence\": \"high\" | \"medium\" | \"low\", \"base\": <the number of respondents it rests on>}],\n \"themes\": [{\"name\": \"<theme in the open-text answers>\", \"respondents\": [\"<every respondent ID whose answer expresses this theme>\"]}],\n \"caveats\": [\"<limitations the decision-makers must know>\"],\n \"recommendation\": \"<what to do, given the decision>\",\n \"would_change_decision\": \"<what evidence would change the recommendation>\",\n \"next_steps\": [{\"action\": \"<what to do>\", \"type\": \"decision\" | \"experiment\" | \"research\" | \"monitor\", \"why\": \"<the finding or gap it answers>\", \"success_measure\": \"<how we will know>\"}]}\nGive percentages as numbers of percent (23.5 for 23.5%), margins of error in percentage points, and yes/no answers as 1 for yes and 0 for no. An answer may belong to more than one theme."
 },
 "outputs": {
  "survey": {
   "screener": [
    {
     "id": "S1",
     "text": "Have you personally cancelled a Fernway meal-kit subscription in the last six months?",
     "type": "single",
     "options": [
      "Yes, I cancelled a Fernway subscription in the last six months",
      "No, I cancelled more than six months ago",
      "No, I am still subscribed to Fernway",
      "No, I have never subscribed to Fernway"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [],
     "show_if": null,
     "qualify": [
      "Yes, I cancelled a Fernway subscription in the last six months"
     ],
     "purpose": "screener"
    },
    {
     "id": "S2",
     "text": "Who mainly made decisions about that Fernway subscription, such as choosing meals, skipping weeks or cancelling?",
     "type": "single",
     "options": [
      "Me alone",
      "Me jointly with someone else",
      "Someone else alone"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [],
     "show_if": null,
     "qualify": [
      "Me alone",
      "Me jointly with someone else"
     ],
     "purpose": "screener"
    },
    {
     "id": "S3",
     "text": "Are you aged 18 or over?",
     "type": "single",
     "options": [
      "Yes",
      "No"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [],
     "show_if": null,
     "qualify": [
      "Yes"
     ],
     "purpose": "screener"
    }
   ],
   "questions": [
    {
     "id": "Q1",
     "text": "In your own words, what was the main reason you cancelled your Fernway subscription?",
     "type": "open",
     "options": [],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [
      "O1"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q2",
     "text": "Which of the following, if any, contributed to your decision to cancel your Fernway subscription? Please select all that apply.",
     "type": "multi",
     "options": [
      "Delivery times or delivery reliability did not suit me",
      "I could not easily skip or pause weeks when I did not want a box",
      "I did not like the taste or quality of the food",
      "I switched to another meal-kit provider",
      "I wanted a break and may return later",
      "My circumstances changed (for example my household, routine or moving home)",
      "The portion sizes did not suit my household",
      "The price was too high or not good value for money",
      "The recipes did not suit my tastes or dietary needs",
      "There was not enough choice or variety of recipes",
      "There were too many meals for my household in some weeks",
      "Another reason"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [
      "O1"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q3",
     "text": "Which one of the following was the single most important reason you cancelled your Fernway subscription? Please select one.",
     "type": "single",
     "options": [
      "Delivery times or delivery reliability did not suit me",
      "I could not easily skip or pause weeks when I did not want a box",
      "I did not like the taste or quality of the food",
      "I switched to another meal-kit provider",
      "I wanted a break and may return later",
      "My circumstances changed (for example my household, routine or moving home)",
      "The portion sizes did not suit my household",
      "The price was too high or not good value for money",
      "The recipes did not suit my tastes or dietary needs",
      "There was not enough choice or variety of recipes",
      "There were too many meals for my household in some weeks",
      "Another reason"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [
      "O1"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q4",
     "text": "Overall, how satisfied or dissatisfied were you with Fernway while you were subscribed?",
     "type": "scale",
     "options": [],
     "scale_points": 5,
     "scale_low": "Very dissatisfied",
     "scale_high": "Very satisfied",
     "objectives": [
      "O4"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q5",
     "text": "To show you are reading the questions carefully, please select \"Rarely\" below.",
     "type": "single",
     "options": [
      "Never",
      "Rarely",
      "Sometimes",
      "Often",
      "Always"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [],
     "show_if": null,
     "qualify": [],
     "purpose": "attention"
    },
    {
     "id": "Q6",
     "text": "How important would each of the following be in deciding whether to subscribe to Fernway again? Please rank them from 1 (most important) to 5 (least important).",
     "type": "ranking",
     "options": [
      "A lower price for my plan",
      "Being able to choose fewer meals in some weeks",
      "Being able to pause my subscription for several weeks at a time",
      "Being able to skip any week with one tap, without having to change my delivery six days ahead",
      "More choice of recipes each week"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [
      "O2"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q7",
     "text": "Thinking back to when you cancelled, if Fernway had offered you just one of these two changes, which one, if either, would have been most likely to make you stay subscribed?",
     "type": "single",
     "options": [
      "A 20% lower price for my plan, with the current rules for skipping a week",
      "The ability to skip any week with one tap in the app, at the same price as I was paying",
      "Neither - I would still have cancelled",
      "Don't know"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [
      "O2",
      "O3"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q8",
     "text": "If nothing changed about Fernway's prices or ability to skip weeks, how likely or unlikely would you be to subscribe to Fernway again in the next six months?",
     "type": "scale",
     "options": [],
     "scale_points": 5,
     "scale_low": "Very unlikely",
     "scale_high": "Very likely",
     "objectives": [
      "O3"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q9",
     "text": "If the price of your previous Fernway plan were 20% lower, and everything else stayed the same, how likely or unlikely would you be to subscribe to Fernway again in the next six months?",
     "type": "scale",
     "options": [],
     "scale_points": 5,
     "scale_low": "Very unlikely",
     "scale_high": "Very likely",
     "objectives": [
      "O2",
      "O3"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q10",
     "text": "If you could skip any week with one tap in the Fernway app, rather than having to change your delivery at least six days ahead, and the price and everything else stayed the same, how likely or unlikely would you be to subscribe to Fernway again in the next six months?",
     "type": "scale",
     "options": [],
     "scale_points": 5,
     "scale_low": "Very unlikely",
     "scale_high": "Very likely",
     "objectives": [
      "O2",
      "O3"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q11",
     "text": "Which of the following best describes what you mainly do now for the meals you previously used Fernway for?",
     "type": "single",
     "options": [
      "I subscribe to another meal-kit service",
      "I sometimes use another meal-kit service but do not subscribe regularly",
      "I cook meals myself without a meal kit",
      "I eat out or get takeaways",
      "Something else"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [
      "O3"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "core"
    },
    {
     "id": "Q12",
     "text": "For how long were you subscribed to Fernway in total before you cancelled?",
     "type": "single",
     "options": [
      "Less than 3 months",
      "3 to 12 months",
      "More than 12 months",
      "Don't know / can't remember"
     ],
     "scale_points": null,
     "scale_low": null,
     "scale_high": null,
     "objectives": [
      "O1",
      "O3"
     ],
     "show_if": null,
     "qualify": [],
     "purpose": "profile"
    }
   ],
   "sample_plan": "Email a unique, single-use survey link to every customer on Fernway's customer database who cancelled a subscription in the last six months (Fernway delivers only in the UK, so no country screening is needed), with reminder emails targeted to under-filled groups. Use database tenure at cancellation to stratify fieldwork and set interlocking quotas of 385 valid completes in each tenure group - under 3 months, 3-12 months and over 12 months - for 1,155 valid completes in total, since 385 per group gives a maximum margin of error of plus or minus 5 percentage points at 95% confidence for a percentage within each group (p=0.5). Note that about 600 completes in total (about 200 per group) would only give approximately plus or minus 7 percentage points per tenure group and therefore would not meet the board's stated precision, so the larger sample is required. Keep fieldwork open for a group until its quota is met, inviting everyone in smaller strata. Append plan type (Couples or Family) and database tenure to each response via the unique link to validate Q12 and for weighting to the known tenure and plan profile of all cancellers in the period. Exclude from valid completes anyone failing S1-S3, failing the Q5 attention check, or speeding/straight-lining, and replace them within quota. Randomise the order of options in Q2, Q3 and Q6 (keeping 'Another reason' last where applicable) and rotate the order of Q9 and Q10 across respondents so neither price nor flexibility is systematically advantaged.",
   "analysis_plan": [
    {
     "objective": "O1",
     "approach": "Code the unaided answers to Q1 into a coding frame built from the responses themselves and report them alongside, but before, prompted results to check prompting effects. Report the percentage selecting each contributing reason at Q2 and each single most important reason at Q3 overall and separately for the three tenure groups from Q12 validated against database tenure, with 95% confidence intervals, and test tenure differences in the Q3 distribution (chi-square, with pairwise comparisons of key reasons including price and inability to skip or pause)."
    },
    {
     "objective": "O2",
     "approach": "Compare flexibility with price three ways: percentage ranking each item first and mean rank at Q6 for lower price versus one-tap skipping, pausing and fewer meals; the head-to-head split at Q7 between a 20% price cut, one-tap skipping and neither; and within-respondent change in likelihood to resubscribe from the no-change baseline Q8 to Q9 (20% price cut) versus to Q10 (one-tap skipping), comparing top-two-box percentages and means. Break each comparison by tenure (Q12) and by single most important cancellation reason (Q3)."
    },
    {
     "objective": "O3",
     "approach": "Define potentially winnable customers as those likely or very likely to resubscribe at Q9 or Q10, and incremental win-back as those likely at Q9 or Q10 but not at the Q8 baseline. Estimate the size of each group, weighted to all cancellers, and identify which change wins them using Q7. Profile winnable versus non-winnable customers by tenure (Q12 and database tenure), plan type from the database, satisfaction (Q4), most important reason for cancelling (Q3) and current alternative (Q11)."
    },
    {
     "objective": "O4",
     "approach": "Report the full distribution, mean score and percentage satisfied (top two points) on Q4 overall, by tenure group (Q12), by most important reason for cancelling (Q3) and for winnable versus non-winnable customers as defined under O3, to show whether dissatisfaction with the product itself, rather than price or flexibility, limits win-back."
    }
   ]
  },
  "analysis": {
   "answers": {
    "price_share": 21.6,
    "price_weighted": 26.3,
    "pause_inflex": 58.6,
    "variety_mentions": 40.0,
    "price_margin": 3.19,
    "family_gap_significant": 0.0,
    "skip_cancellers": 3193.0,
    "variety_long_tenure": 25.0,
    "skip_vs_price_gap": 35.7,
    "completion_rate": 6.5,
    "price_cut_price_leavers": 49.3
   },
   "findings": [
    {
     "statement": "Do not answer the chief executive with an unweighted yes: price is important, especially for early leavers, but difficulty skipping is still the larger stated reason after correcting the sample skew.",
     "evidence": "Unweighted Q1: hard to skip 35.5% versus price 21.6% (base 640; price 95% margin of error +-3.2pp). Reweighted to the 9,850 canceller tenure mix (45% under 3 months, 35% 3-12, 20% over 12), price rises to 26.3% and hard to skip falls to 32.4% (about 3,193 cancellers). The gap narrows from 13.9pp to about 6.1pp, so the unweighted headline overstates the dominance of skipping, but does not reverse the ranking.",
     "confidence": "medium",
     "base": 640
    },
    {
     "statement": "One-tap skipping has the stronger stated win-back pull overall, and particularly among the segment it is meant to fix.",
     "evidence": "Net likely to return overall: one-tap skip 36.4% versus 20% cheaper 25.8% (base 640). Among hard-to-skip leavers (base 227), 58.6% say likely with one-tap skipping versus 22.9% with a price cut, a 35.7pp gap. Among price leavers (base 138), the pattern reverses sensibly: 49.3% likely with a cut versus 24.6% with skipping. This internal consistency supports a real segment-specific demand, though stated likelihood is not observed rejoining.",
     "confidence": "medium",
     "base": 640
    },
    {
     "statement": "A blanket 20% price cut is a blunt and expensive response: it speaks mainly to the early-tenure price segment, while most leavers say it would not bring them back.",
     "evidence": "Only 25.8% overall say likely to return at 20% less, and even among price leavers only 49.3% say likely (base 138). Price as main reason falls with tenure: 31.2% under 3 months (base 138), 27.2% at 3-12 months (base 195), 13.7% over 12 months (base 307). The survey contains no margin, willingness-to-pay, current-customer or incremental-revenue evidence to justify giving 20% to everyone, including customers who would stay anyway.",
     "confidence": "medium",
     "base": 640
    },
    {
     "statement": "The Family plan is not shown to be the problem: the apparent satisfaction gap is small, not statistically significant, and reverses once tenure is accounted for.",
     "evidence": "Satisfied (4-5): Family 49.2% (base 301) versus Couples 53.4% (base 339), a 4.2pp gap against an approximate 95% margin of error for the difference of about +-7.8pp, so not significant at 95%. Within tenure, Family is slightly higher in every band: under 3 months 69.2% (base 26) versus 66.1% (base 112); 3-12 months 61.6% (base 73) versus 57.4% (base 122); over 12 months 42.1% (base 202) versus 35.2% (base 105). Family respondents are disproportionately long-tenure (202 of 301 over 12 months versus 105 of 339 Couples), and satisfaction falls sharply with tenure. This is a tenure mix effect (Simpson's paradox), not evidence of a Family-plan defect, though the smallest cell (base 26) is imprecise.",
     "confidence": "high",
     "base": 640
    },
    {
     "statement": "Long-tenure fatigue, especially recipe repetition, is a separate retention problem that neither skipping nor a price cut fully solves.",
     "evidence": "Satisfied (4-5) falls from 66.7% under 3 months (base 138) to 59.0% at 3-12 months (base 195) to 39.7% over 12 months (base 307). 'Recipes got repetitive' as main reason rises from 6.5% to 14.4% to 25.4% across the same tenure bands. In open text, 40 of 190 answers mention variety or repetition, 25 of them from over-12-month customers. Repetitive-recipe leavers (base 115) are only 28.7% likely to return with skipping and 19.1% with a price cut, so a menu-rotation response is still needed.",
     "confidence": "high",
     "base": 640
    },
    {
     "statement": "Northern Ireland shows a delivery warning signal that must be checked in operational data immediately, but the survey alone cannot size it.",
     "evidence": "66.7% of Northern Ireland respondents cite delivery problems as their main reason (base 12, i.e. about 8 people), versus 8.1% in England (base 546), 13.0% in Scotland (base 46) and 0.0% in Wales (base 36). With base 12, the approximate 95% margin of error is about +-27pp, so the true share among NI cancellers who would respond could be far lower or higher, and aggrieved customers may have been more likely to respond. The signal is too large to ignore and too imprecise to fund a major fix from the survey alone; one NI open-text answer (R445) also complains about promised-day delivery and missing items.",
     "confidence": "low",
     "base": 12
    },
    {
     "statement": "The open-text demand labelled 'skipping' is broader than one tap on a single week: customers also want multi-week pauses, a later cut-off, restored settings, fewer meals per week and less frequent boxes.",
     "evidence": "Of 190 open answers, coded with multiple themes allowed: 47 mention easy skip/pause/hold, 32 mention fewer or variable meals per week or less frequent delivery (for example two-meal weeks, fortnightly boxes), 42 mention price/value, 40 variety/repetition, 19 packaging, 18 delivery or missing/damaged items, 12 portion size and 6 say nothing/life circumstances. Building only a literal one-week skip without pause length, deadline and volume flexibility risks missing the underlying need: too much food arriving too often for unpredictable lives.",
     "confidence": "medium",
     "base": 190
    },
    {
     "statement": "Segments who left for delivery, packaging, portions or repetition are largely unmoved by either offer, so the binary choice will not win back most cancellers.",
     "evidence": "Net likely to return with one-tap skipping / with 20% less respectively: delivery-problem leavers 17.2% / 22.4% (base 58); packaging leavers 12.9% / 9.7% (base 31); portion leavers 27.9% / 11.6% (base 43); repetitive-recipe leavers 28.7% / 19.1% (base 115). Even overall, 63.6% are not net likely to return for skipping and 74.2% are not net likely for the price cut (base 640). Expectations for win-back volume must be set well below the headline 'likely' figures.",
     "confidence": "high",
     "base": 640
    }
   ],
   "themes": [
    {
     "name": "Price, value for money or discounts",
     "respondents": [
      "R041",
      "R054",
      "R074",
      "R077",
      "R099",
      "R103",
      "R104",
      "R120",
      "R123",
      "R124",
      "R154",
      "R158",
      "R181",
      "R194",
      "R197",
      "R204",
      "R225",
      "R251",
      "R260",
      "R272",
      "R286",
      "R296",
      "R325",
      "R328",
      "R331",
      "R336",
      "R340",
      "R348",
      "R356",
      "R380",
      "R426",
      "R430",
      "R436",
      "R437",
      "R447",
      "R449",
      "R454",
      "R483",
      "R492",
      "R495",
      "R500",
      "R562"
     ]
    },
    {
     "name": "Easy skipping, pausing or holding the subscription",
     "respondents": [
      "R031",
      "R033",
      "R040",
      "R048",
      "R049",
      "R074",
      "R084",
      "R089",
      "R093",
      "R121",
      "R123",
      "R133",
      "R179",
      "R185",
      "R186",
      "R197",
      "R217",
      "R220",
      "R245",
      "R264",
      "R295",
      "R303",
      "R308",
      "R316",
      "R351",
      "R372",
      "R376",
      "R380",
      "R385",
      "R393",
      "R410",
      "R425",
      "R433",
      "R434",
      "R447",
      "R480",
      "R493",
      "R527",
      "R530",
      "R542",
      "R559",
      "R562",
      "R572",
      "R590",
      "R599",
      "R603",
      "R613"
     ]
    },
    {
     "name": "Fewer meals per week, variable meal numbers or less frequent boxes",
     "respondents": [
      "R007",
      "R011",
      "R031",
      "R059",
      "R063",
      "R080",
      "R098",
      "R103",
      "R121",
      "R138",
      "R165",
      "R180",
      "R206",
      "R222",
      "R232",
      "R246",
      "R257",
      "R270",
      "R327",
      "R328",
      "R337",
      "R342",
      "R361",
      "R373",
      "R408",
      "R422",
      "R436",
      "R459",
      "R502",
      "R540",
      "R619",
      "R634"
     ]
    },
    {
     "name": "Recipe variety or repetitive recipes",
     "respondents": [
      "R014",
      "R028",
      "R038",
      "R089",
      "R100",
      "R115",
      "R130",
      "R133",
      "R148",
      "R164",
      "R180",
      "R186",
      "R200",
      "R215",
      "R237",
      "R242",
      "R254",
      "R257",
      "R283",
      "R312",
      "R316",
      "R367",
      "R377",
      "R382",
      "R396",
      "R407",
      "R408",
      "R412",
      "R416",
      "R419",
      "R425",
      "R524",
      "R525",
      "R529",
      "R539",
      "R560",
      "R589",
      "R608",
      "R614",
      "R615"
     ]
    },
    {
     "name": "Portion sizes too small or not adjustable",
     "respondents": [
      "R021",
      "R030",
      "R107",
      "R134",
      "R137",
      "R379",
      "R401",
      "R402",
      "R535",
      "R624",
      "R629",
      "R632"
     ]
    },
    {
     "name": "Delivery reliability, timing, missing or damaged items",
     "respondents": [
      "R012",
      "R058",
      "R068",
      "R070",
      "R163",
      "R170",
      "R193",
      "R247",
      "R298",
      "R311",
      "R349",
      "R359",
      "R414",
      "R445",
      "R460",
      "R551",
      "R609",
      "R630"
     ]
    },
    {
     "name": "Packaging waste, recyclability or packaging condition",
     "respondents": [
      "R051",
      "R053",
      "R057",
      "R058",
      "R140",
      "R163",
      "R199",
      "R247",
      "R262",
      "R280",
      "R298",
      "R319",
      "R414",
      "R439",
      "R445",
      "R487",
      "R551",
      "R622",
      "R640"
     ]
    },
    {
     "name": "Nothing would have helped or life circumstances",
     "respondents": [
      "R096",
      "R357",
      "R429",
      "R523",
      "R621",
      "R628"
     ]
    }
   ],
   "caveats": [
    "Only 640 of 9,850 invited cancellers completed the survey (6.5%; 702 started), so non-response bias could substantially change every estimate. The completes are demonstrably unrepresentative: over 12 months tenure is 48.0% of the sample versus 20% of all cancellers, under 3 months is 21.6% versus 45%, and Family plan is 47.0% versus 34%. All supplied tables are unweighted; tenure-reweighted figures in this analysis correct only that one known skew and cannot correct for unobserved differences between responders and non-responders.",
    "Respondents are cancellers only, surveyed in September 2025 about cancellations from March to August 2025. There are no current customers, prospects or never-customers, so evidence about keeping current customers or market-level price elasticity is indirect. Retrospective satisfaction and reasons given after leaving may be coloured by the cancellation itself and by recall over up to about six months.",
    "Q3 and Q4 are hypothetical stated intentions to come back, asked in a fixed order (skipping before price), with no cost, effort, competitor or timing specified. Stated 'fairly likely' in particular usually overstates actual behaviour. They should be used to compare relative pull within this sample, not forecast reactivation volumes or revenue. No combined option (skip plus smaller cut), no smaller-price-cut ladder and no fewer-meals option were tested, even though open text suggests volume flexibility may be part of the need.",
    "Q1 forces a single main reason, which understates multi-causal churn. Open answers repeatedly combine causes (for example pause plus price, or repetition plus ease of skipping), and only 190 of 640 respondents (29.7%) gave an open answer, so open-text counts indicate what themes exist and how they are described, not population shares.",
    "Subgroup bases are small where stakeholders most want certainty: Northern Ireland base 12, Wales 36, Scotland 46, packaging leavers 31, other leavers 28, and under-3-months Family satisfaction base 26. Differences involving these cells have wide margins of error and should trigger checks in administrative data rather than direct investment.",
    "Tenure, plan and region in the survey appear to be self-reported profile answers and were not shown linked to the customer database. Misreporting and joint weighting (tenure by plan by region jointly, not tenure alone) were not available from the supplied tables. The median completion time of 5 minutes 20 seconds and the 62 starters who did not complete give no evidence about careless responding or differential drop-out, which were not analysed.",
    "The survey cannot evaluate the economics of either option: margin per box, delivery and ingredient cost savings from skipped weeks, cannibalisation if existing customers skip more paid weeks, effect of a price cut on existing-customer revenue, promotional expectations created, or delivery-network costs in Northern Ireland. A preference ranking is not a business case."
   ],
   "recommendation": "Prioritise flexibility over a blanket price cut: commit to building easy skipping/pausing rather than cutting all plan prices by 20%, but scope it as the flexibility customers actually describe (one-tap skip, multi-week pause/hold with settings retained, and a later, clearer cut-off), and put a cheap behavioural test and CRM/operations validation in front of the full four-month build. In parallel, check Northern Ireland delivery in operational data this week, and treat long-tenure recipe fatigue as a separate workstream. Do not spend the margin of a permanent 20% cut on the basis of this survey: weighted price churn (26.3%) remains second to skipping (32.4%), only 49.3% of even price leavers say a cut would likely bring them back, and there is no evidence here on current customers or incremental profit.",
   "next_steps": [
    {
     "action": "Decide now: approve flexibility (skip/pause) as the lead investment and reject a permanent across-the-board 20% price cut; reserve any discounting for targeted, tested save and win-back offers.",
     "type": "decision",
     "why": "Weighted Q1 and Q3/Q4 favour skipping, while a blanket cut lacks economic evidence and would subsidise customers who would stay anyway.",
     "success_measure": "A recorded decision with an agreed scope (single-week skip, multi-week pause, cut-off timing, settings retention) and an explicit non-decision on list price pending experiment results."
    },
    {
     "action": "Within days, validate and reweight against the customer database: link cancellations to tenure, plan, region, price paid, discount expiry, skip attempts/failures, unwanted deliveries, holiday-season timing and delivery performance; produce joint tenure-by-plan weighted estimates for the survey.",
     "type": "research",
     "why": "The sample is skewed on tenure and plan, the Family gap is a mix effect, and Northern Ireland rests on base 12. Administrative data can confirm or refute each survey signal at full-population scale.",
     "success_measure": "Full-population rates for all 9,850 cancellers and comparable current customers, including Northern Ireland on-time, missing-item and damaged-box rates versus other nations, with confidence intervals and a clear confirm/refute verdict per stakeholder question."
    },
    {
     "action": "Audit Northern Ireland delivery operations now (carrier, depot, promised versus actual day/slot, missing items, refunds and contacts), and fix any confirmed defect before using NI in marketing or win-back.",
     "type": "monitor",
     "why": "66.7% of NI respondents (base 12) cite delivery versus 8.1% in England; imprecise, but the downside of a real regional failure is high and the operational check is cheap.",
     "success_measure": "NI delivery KPIs reported weekly and brought within an agreed tolerance of the UK average (for example on-time-in-full gap closed), or documented evidence that no operational gap exists and the survey signal was a small-base artefact."
    },
    {
     "action": "Run a randomised behavioural test before or alongside the full build: offer eligible recent cancellers and at-risk current customers (a) easy skip/pause (concierge or minimum viable feature), (b) a 20% discount, (c) both if feasible, and (d) control, measuring actual behaviour not intentions.",
     "type": "experiment",
     "why": "Q3/Q4 are hypothetical, fixed-order stated preferences. The decision needs incremental reactivation, retention and margin per arm, including whether skipping cannibalises paid weeks.",
     "success_measure": "Pre-registered primary measures at 12 and 26 weeks: rejoin or save rate, active weeks retained, boxes skipped per retained customer, revenue and contribution margin per customer versus control, by tenure, plan and original cancellation reason."
    },
    {
     "action": "Prototype the exact flexibility proposition with recent cancellers and current customers: one-tap skip, pause length and limits, cut-off deadline, reminders, restoring preferences, plus two-meal and fortnightly variants emerging from open text.",
     "type": "research",
     "why": "Open text (47 skip/pause mentions; 32 fewer/variable meals or less frequent boxes, base 190) suggests a literal one-week skip may underspecify the need. Requirements must be settled before four months of build.",
     "success_measure": "Task success and comprehension in usability testing (for example at least 90% can skip and pause for a holiday unaided), plus ranked preference among skip, pause, two-meal and fortnightly options by tenure segment."
    },
    {
     "action": "Create a long-tenure menu-variety workstream: expand and evidence rotation depth, reduce repeat frequency within a rolling window for long-tenure customers, and test new-cuisine slots with over-12-month customers.",
     "type": "experiment",
     "why": "Repetitive recipes rise to 25.4% of over-12-month cancellers (base 307), satisfaction falls to 39.7% in that group, and 25 of 40 open-text variety mentions come from over-12-month customers; neither skipping nor price moves this segment much.",
     "success_measure": "Repeat exposure per customer over rolling 8-12 weeks, menu-choice satisfaction, skip-for-no-choice rate, and 26-week retention among over-12-month customers in test versus control."
    },
    {
     "action": "After launch, monitor a cancellation-reason and save-flow dashboard weighted to the canceller population, with minimum-base rules for reporting small nations and segments.",
     "type": "monitor",
     "why": "Unweighted, small-base reporting created the misleading Family and Northern Ireland readings in this survey; ongoing decisions need representative, precision-aware tracking.",
     "success_measure": "Monthly weighted churn reasons with bases and margins of error published alongside every percentage, skip/pause usage before cancellation, save rate after a skip/pause offer, and no subgroup reported without its base."
    }
   ],
   "would_change_decision": "The recommendation would change if behavioural evidence contradicted stated preference: for example, a randomised win-back or save-flow experiment showing a 20% (or smaller targeted) discount produces materially higher incremental retained margin per contacted canceller or at-risk customer than easy skipping; CRM data showing most cancellations follow price-rise or discount-expiry events rather than unwanted deliveries, failed skips or holiday periods; evidence from current (not only cancelled) customers that price is their dominant imminent churn driver; a prototype showing the real unmet need is a two-meal or fortnightly plan rather than skipping, so volume flexibility should be built instead of or first; or financial modelling showing easier skipping mainly removes paid weeks from customers who would have stayed anyway, reducing lifetime value more than it extends tenure."
  },
  "score": {
   "task_id": "sv-001",
   "sound": false,
   "research_score": 90.66,
   "plan": {
    "produced": true,
    "questions": 12,
    "flaws": {},
    "question_flaws": {},
    "ambiguous": [],
    "screener_ok": true,
    "objectives_answerable": 1,
    "hypothesis_tested": true,
    "precision_handled": false,
    "questions_clean": 1.0,
    "critical_checks": 0,
    "failures": [
     "precision"
    ],
    "sound": false,
    "points": 90.0
   },
   "analysis": {
    "produced": true,
    "numeric_correct": {
     "price_share": true,
     "price_weighted": true,
     "pause_inflex": true,
     "variety_mentions": true,
     "price_margin": true,
     "family_gap_significant": true,
     "skip_cancellers": true,
     "variety_long_tenure": true,
     "skip_vs_price_gap": true,
     "completion_rate": true,
     "price_cut_price_leavers": true
    },
    "numeric_given": {
     "price_share": 21.6,
     "price_weighted": 26.3,
     "pause_inflex": 58.6,
     "variety_mentions": 40.0,
     "price_margin": 3.19,
     "family_gap_significant": 0.0,
     "skip_cancellers": 3193.0,
     "variety_long_tenure": 25.0,
     "skip_vs_price_gap": 35.7,
     "completion_rate": 6.5,
     "price_cut_price_leavers": 49.3
    },
    "findings_conveyed": 1,
    "traps_handled": {
     "ni_delivery": true,
     "family_reversal": true,
     "tenure_skew": true,
     "stated_intent": true,
     "price_hypothesis": true,
     "nonresponse": true,
     "no_comparison": true,
     "question_order": true
    },
    "material_unsupported": 0,
    "recommendation_ok": true,
    "theme_recall": 1.0,
    "theme_precision": 0.9890635680109364,
    "right_steps": {
     "behaviour_test": true,
     "ni_evidence": true,
     "weighting": true,
     "price_new_customers": false,
     "stayers_comparison": true
    },
    "wrong_steps": [],
    "confidence_ok": false,
    "decision_criteria_ok": true,
    "failures": [],
    "sound": true,
    "points": 90.95
   },
   "cost_usd": null,
   "latency_ms": 269191,
   "errors": {}
  }
 }
}
