Benchmarks / SurveyBench / Why Fernway customers cancel / Gemini 4 Argon

Measured by Spring Prompt

Gemini 4 Argon: Why Fernway customers cancel

The decision: 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. The brief and the data →

Verdict
✗ Not sound
Research score
82 of 100
Analysis rating
985
Head to head, this task
won 14 of 36

Why it is not sound

The survey

  • A double-barrelled question: Q3

The analysis

  • Reported a material finding the data does not support
  • Missed a trap: Only customers who cancelled were surveyed, so the survey cannot show what distinguishes leavers from customers who stay; the reasons are self-reported.
  • Missed a trap: The one-tap skipping question (Q3) came immediately before the price-cut question (Q4), so answers to Q4 may have been framed by Q3.

Findings the judges found unsupported

  • One-tap skipping is a much stronger win-back lever than a 20% price cut, both overall and when weighted to the true population, without sacrificing 20% of revenue.The supplied tables do not establish the overall win-back comparison after weighting for tenure: Q3 and Q4 are not cross-tabulated by tenure. The survey supports stronger stated interest in skipping, not a demonstrated stronger commercial win-back effect.
  • Open-text responses reveal an unmeasured product gap: 'difficulty skipping' is heavily intertwined with box-size inflexibility (minimum 3-4 meals per week) and multi-week holiday pausing. Many customers want a 2-meal option or fortnightly cadence.The reported count of 35 skipping/pausing answers understates the supplied open-text evidence and conflicts with the analysis's own later count of 47. The broader flexibility theme is supported, but this quantitative evidence is not.
  • Recipe fatigue is the silent killer of long-term retention and Family plan subscribers, driving declining satisfaction over time.The cross-sectional tables do not establish that recipe fatigue causes satisfaction to decline over time or is the main reason for that decline. The Family-plan difference in repetition complaints is not adjusted for tenure. Recipe variety is also not the largest open-text theme: price and skipping/pausing have more mentions, even using the researcher's own final counts.

Scorecard

Numbers 11 of 11 right

QuestionIts answerTrue
What percentage of respondents gave price as their main reason for cancelling?✓21.6%21.6%
Weighted to the tenure mix of all customers who cancelled, what percentage would give price as their main reason?✓26.3%26.3%
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?✓58.6%58.6%
How many of the open-text answers mention recipe variety or repetitive recipes?✓3940
What is the 95% margin of error, in percentage points, on the unweighted share of respondents giving price as their main reason?✓3.2 points3.19 points
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)✓NoNo
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?✓3,1933,193
How many of the open-text answers from respondents who were customers for over 12 months mention recipe variety or repetitive recipes?✓2425
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?✓35.7 points35.7 points
What percentage of the customers invited completed the survey?✓6.5%6.5%
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?✓49.3%49.3%

Traps in the data 6 of 8 handled

  • ✓Northern Ireland respondents (base 12) cite delivery problems far more often.
  • ✓Family-plan respondents look less satisfied overall, but are more satisfied than Couples within every tenure band: Family customers skew to long tenure, and long-tenure customers are less satisfied.
  • ✓Long-tenure customers are 50% of respondents but 20% of cancellers; short-tenure customers, who cite price more, are under-represented.
  • ✓'Likely to come back' answers are stated intentions about a hypothetical feature.
  • ✓The chief executive believes customers leave mainly because of price.
  • ✓Only 702 of 9,850 invited customers started the survey (about 7%); those who respond may differ from those who do not.
  • ✗Only customers who cancelled were surveyed, so the survey cannot show what distinguishes leavers from customers who stay; the reasons are self-reported.
  • ✗The one-tap skipping question (Q3) came immediately before the price-cut question (Q4), so answers to Q4 may have been framed by Q3.

The analysis

Recommendation

Build the one-tap skipping feature rather than cutting prices by 20%, and expand the four-month product scope to include multi-week pausing ('holiday hold'), bug fixes for the mobile app skip button, and a later cut-off window. Do not cut plan prices by 20%: it addresses a smaller share of churn, shows lower win-back potential (25.8% vs. 36.4%), and destroys margin across the entire active base. Alongside building one-tap skip, test a 2-meal-per-week plan and targeted retention discounts for price-sensitive early tenure (<3m) customers.

What would change it

Financial modelling and live A/B testing showing that one-tap skipping increases skip frequency among active, non-churning subscribers by >20% (cannibalising weekly order volume by more than the value of retained subscribers), OR win-back experiments showing that churned customers who claimed they would return for one-tap skipping fail to reactivate in practice while targeted 15-20% discounts drive profitable reactivation.

Findings

  1. To answer the CEO: Price is not the primary driver of churn—inflexibility in skipping weeks is. Even after weighting the survey to correct for the under-representation of short-tenure customers (who are more price-sensitive), 'hard to skip weeks' remains the #1 reason for leaving.

    High confidenceBase 640

    Unweighted, 35.5% left due to difficulty skipping weeks vs. 21.6% for price (n=640). In the database, 45% of cancellers have <3 months tenure (where price churn is 31.2%) vs. 21.6% in our sample. Weighted to true population tenure, 'hard to skip weeks' is still #1 at 32.4% (~3,193 customers) vs. 26.3% (~2,591 customers) for price.

  2. One-tap skipping is a much stronger win-back lever than a 20% price cut, both overall and when weighted to the true population, without sacrificing 20% of revenue.

    High confidenceBase 640

    Overall, 36.4% of respondents are likely to return for one-tap skipping vs. 25.8% for a 20% price cut (n=640). Among those who left over skipping difficulty (n=227), 58.6% would return for one-tap skipping vs. 22.9% for a price cut (a 35.7pp gap). Among price leavers (n=138), 49.3% would return for a price cut vs. 24.6% for one-tap skipping.

  3. To answer the Head of Product: The Family plan does not have a satisfaction problem. The apparent lower satisfaction is a statistical illusion (Simpson's paradox) caused by tenure mix: long-tenure customers are less satisfied, and Family plan respondents in our sample are heavily skewed toward long tenure.

    High confidenceBase 640

    Overall Family satisfaction (49.2%, n=301) is slightly lower than Couples (53.4%, n=339), a statistically insignificant difference (p=0.29). However, within every tenure band, Family satisfaction is actually higher than Couples: <3 months (69.2% vs 66.1%), 3-12 months (61.6% vs 57.4%), and >12 months (42.1% vs 35.2%). Family looks worse overall only because 67.1% of Family respondents had >12 months tenure (where satisfaction is 39.7%), compared to 31.0% of Couples.

  4. To answer the Operations Director: The survey suggests a severe delivery problem in Northern Ireland, where two-thirds of respondents cited delivery as their main reason for cancelling, but the sample is far too small to be conclusive on its own.

    Low confidenceBase 12

    66.7% (8 of 12) of Northern Ireland respondents cancelled due to delivery problems, compared to 8.1% in England (n=546), 13.0% in Scotland (n=46), and 0.0% in Wales (n=36). Although statistically significant vs. the rest of the UK, a base of 12 is too small to rely on without checking operational courier logs.

  5. Open-text responses reveal an unmeasured product gap: 'difficulty skipping' is heavily intertwined with box-size inflexibility (minimum 3-4 meals per week) and multi-week holiday pausing. Many customers want a 2-meal option or fortnightly cadence.

    Medium confidenceBase 190

    Of 190 open-text responses, 35 mention skipping/pausing (including multi-week holiday pauses, app crashes, and early cut-off deadlines), while 29 spontaneously state they would have stayed if they could order fewer meals per week (e.g., 2 meals instead of a 3-meal minimum).

  6. Recipe fatigue is the silent killer of long-term retention and Family plan subscribers, driving declining satisfaction over time.

    High confidenceBase 640

    Menu repetition is the #3 reason for cancelling overall (18.0%), but jumps from 6.5% among <3 month customers (n=138) to 25.4% among >12 month customers (n=307), and is much higher on the Family plan (23.6%, n=301) than Couples (13.0%, n=339). It is the single largest theme in open-text feedback (39 of 190 answers, 24 of which are from >12 month customers).

Caveats

  • Severe sample skew: Long-tenure (>12m) customers are 48.0% of the sample vs. 20.0% of the cancelled population, and Family plan customers are 47.0% vs. 34.0%. Because long-tenure customers are less price-sensitive and more fatigued by recipes, unweighted totals understate price sensitivity (21.6% unweighted vs. 26.3% weighted) and overstate recipe repetition.
  • Non-response bias: With a 6.5% completion rate (640 of 9,850), more engaged or frustrated long-term customers were much more likely to respond than early churners who left after an introductory discount.
  • Stated preference inflation: Q3 and Q4 measure stated intent to return, which always overstates actual reactivation rates. A 20% across-the-board price cut guarantees an immediate 20% revenue loss on retained customers, whereas stated return rates will only partially convert.
  • Confounded Q1 option: 'Fewer meals per week' (e.g., 2 meals instead of 3+) was not an option in Q1, despite being cited by 29 respondents in Q5. Many of these customers likely selected 'Hard to skip weeks' or 'Price' in Q1 because they were forced to buy more food than they could eat.

Next steps

  • decisionProceed with building one-tap week skipping, ensuring the spec includes multi-week holiday pausing (up to 8 weeks), fixing the mobile app skip crash, and reviewing the weekly cut-off deadline.
  • experimentRun a commercial and operational experiment offering a 2-meal-per-week box option (and/or fortnightly cadence) to customers attempting to cancel.
  • researchAudit courier performance, late deliveries, and missing ingredient rates in Northern Ireland (and Scotland) using internal operations data.
  • experimentTest stepped introductory pricing or targeted loyalty discounts at month 2 and month 6 instead of a blanket 20% price cut.
  • monitorOverhaul menu rotation for customers over 6 months tenure, specifically reducing repetition of chicken traybakes, curries, and pastas.

Open-text themes it coded

Recipe variety and menu repetition 40Price, affordability, and loyalty discounts 42Ease of skipping weeks, pausing, and cut-off deadlines 47Fewer meals per week and smaller box options 30Packaging waste and recyclability 18Delivery reliability, timing, and missing/damaged items 18Portion sizes too small or inflexible 12External life changes (unavoidable churn) 6

The survey it planned

2 screening questions and 12 questions, as the model wrote them.

  1. S1

    What is the current status of your Fernway meal-kit subscription?

    One answer
    • I am currently subscribed and receiving boxes
    • I cancelled my subscription within the last 6 months
    • I cancelled my subscription more than 6 months ago
    • I have never subscribed to Fernway

    Continues if I cancelled my subscription within the last 6 months

  2. S2

    How long were you subscribed to Fernway before you cancelled?

    One answer
    • Less than 3 months
    • 3 to 12 months
    • More than 12 months

    Continues if Less than 3 months; 3 to 12 months; More than 12 months

  3. Q1

    Which Fernway plan were you subscribed to when you cancelled?

    One answer
    • Couples plan (recipes for two)
    • Family plan (recipes for four)
    • Not sure
  4. Q2

    Thinking back to your time with Fernway, how satisfied or dissatisfied were you with your experience overall?

    Scale

    Scale 1-5: Very dissatisfied to Very satisfied

  5. Q3

    How satisfied or dissatisfied were you with the freshness and quality of the ingredients you received?

    Scale

    Scale 1-5: Very dissatisfied to Very satisfied

    Judges: double-barrelled

    Freshness and ingredient quality can receive different satisfaction ratings, but only one answer is allowed.

    Asks about two things at once ('freshness' and 'quality'), which could yield different satisfaction levels.

  6. Q4

    How satisfied or dissatisfied were you with the ease of managing your subscription (such as skipping weeks or changing deliveries) in the app?

    Scale

    Scale 1-5: Very dissatisfied to Very satisfied

    Judges: assumptive

    Assumes respondents managed their subscription in the app. The screeners do not establish app use, and there is no not-applicable option.

    Assumes the respondent tried to manage their subscription specifically 'in the app' and provides no 'Not applicable' option if they did not.

  7. Q5

    Which of the following contributed to your decision to cancel your Fernway subscription? Please select all that apply.

    Any that apply
    • The boxes were too expensive for my budget
    • I had to give too much notice (6 days) to skip a week
    • It was too difficult or inconvenient to skip or pause deliveries in the app
    • I wanted to order fewer meals per week than the plan allowed
    • We were wasting food because we couldn't use all the meals in time
    • Issues with ingredient quality, freshness, or missing items
    • Lack of variety or appeal in the weekly recipes
    • Delivery issues (e.g. late, damaged, or missed boxes)
    • My introductory discount ended
    • My personal or household circumstances changed
    • I switched to another meal-kit provider
    • Other
  8. Q6

    Which SINGLE reason was the most important in your decision to cancel?

    One answer
    • The boxes were too expensive for my budget
    • I had to give too much notice (6 days) to skip a week
    • It was too difficult or inconvenient to skip or pause deliveries in the app
    • I wanted to order fewer meals per week than the plan allowed
    • We were wasting food because we couldn't use all the meals in time
    • Issues with ingredient quality, freshness, or missing items
    • Lack of variety or appeal in the weekly recipes
    • Delivery issues (e.g. late, damaged, or missed boxes)
    • My introductory discount ended
    • My personal or household circumstances changed
    • I switched to another meal-kit provider
    • Other
  9. Q7

    While you were subscribed, did you ever end up paying for and receiving a Fernway box that you did not want because you missed the 6-day cut-off to skip?

    One answer
    • Yes, more than once
    • Yes, once
    • No, never
    • I didn't know I could skip weeks
  10. Q8

    Please rank the following potential improvements to Fernway in order of how appealing they are to you (where 1 is the most appealing and 5 is the least appealing).

    Ranking
    • A permanent 20% reduction in the price of your weekly box
    • One-tap skipping of any delivery week in the app, up to 48 hours before delivery
    • The ability to choose fewer meals per week when you need less food
    • The ability to pause your subscription for up to 3 months with one tap
    • A wider selection of weekly recipes to choose from
  11. Q9

    To make sure our survey is working properly, please select 'Somewhat likely' as your answer to this question.

    One answer
    • Very unlikely
    • Somewhat unlikely
    • Neither likely nor unlikely
    • Somewhat likely
    • Very likely

    Continues if Somewhat likely

  12. Q10

    Imagine Fernway kept its current prices, but introduced one-tap skipping in the app (allowing you to easily skip any week with shorter notice). How likely would you be to restart your subscription?

    Scale

    Scale 1-5: Very unlikely to Very likely

  13. Q11

    Now imagine Fernway kept its current 6-day skip rules, but permanently cut the price of all plans by 20%. How likely would you be to restart your subscription?

    Scale

    Scale 1-5: Very unlikely to Very likely

  14. Q12

    If you had to choose between the two options below, which one would make you MORE likely to rejoin Fernway?

    One answer
    • A permanent 20% price cut on all plans (with the current 6-day skip rule)
    • One-tap skipping of any week in the app (at the current plan prices)
    • Neither of these would make me consider rejoining
Sample plan

Invite all UK customers in the Fernway database who cancelled their subscription within the last 6 months via email. Note on precision constraint: Achieving a +/- 5 percentage point margin of error at 95% confidence for EACH of the three tenure groups (under 3 months, 3-12 months, over 12 months) mathematically requires 385 completes per group (1,155 total completes), unless finite population correction applies to a small total pool of churned users. If strictly capped at 600 total completes, we will set equal quotas of n=200 per tenure group, which yields a margin of error of +/- 6.9 percentage points per group (and +/- 4.0 percentage points overall). We recommend either expanding the sample cap to 1,155 completes (385 per tenure quota) to strictly meet the board's +/- 5% subgroup precision requirement, or applying post-stratification weights to the 600 completes (200 per group) to match the true database tenure distribution for overall reporting. Respondents who fail the attention check (Q9) will be excluded from the final count.