Benchmarks / SurveyBench / Should Corefit gyms stop opening 24 hours? / GPT-6.1 Sol

Measured by Spring Prompt

GPT-6.1 Sol: Should Corefit gyms stop opening 24 hours?

The decision: whether to close all 14 gyms between 11pm and 5am, close overnight only at the quietest gyms, or keep every gym open 24 hours. The brief and the data →

Verdict
✗ Not sound
Research score
98 of 100
Analysis rating
1,650
Head to head, this task
won 9 of 9

Why it is not sound

The analysis

  • Missed a trap: Only current members were surveyed: people who joined for 24-hour access and have since left, and people who would join only because of it, are missing, so the survey cannot show what 24-hour opening does for recruitment or the brand; and night-shift workers are a small group whose answers carry little weight in the totals even though they depend on the hours.

Scorecard

Numbers 11 of 11 right

QuestionIts answerTrue
What percentage of respondents reported at least one visit to a Corefit gym between 11pm and 5am in the last four weeks?✓14.9%14.9%
What percentage of all respondents said they would be likely (fairly or very) to cancel their membership if their home gym closed between 11pm and 5am?✓14.6%14.6%
Among respondents who reported at least one visit between 11pm and 5am in the last four weeks, what percentage said they would be likely (fairly or very) to cancel?✓45.8%45.8%
Weighted so that respondents with at least one visit between 11pm and 5am in the last four weeks count as 6% of members (the door-entry figure) and the rest as 94%, what percentage would be likely (fairly or very) to cancel?✓11.4%11.4%
If the weighted share likely to cancel were applied to all 38,600 members, about how many members would that be?✓4,4004,383
Among respondents who reported at least one visit between 11pm and 5am in the last four weeks, in which band of the Q1 table does the median number of visits fall? Give the lowest value of that band (for example 4 for '4 to 7').✓22
What is the 95% margin of error, in percentage points, on the unweighted share of respondents who reported at least one visit between 11pm and 5am in the last four weeks?✓2.27 points2.27 points
Is the difference between North West England and North East England respondents in the share likely (fairly or very) to cancel statistically significant at the 95% level? (1 for yes, 0 for no)✓NoNo
How many of the open-text answers say that 11pm to 5am is the only time the respondent can train because they work shifts or nights?✓1616
How many of the open-text answers from respondents who visited between 11pm and 5am in the last four weeks say they would move to another gym?✓1919
What percentage of the members invited completed the survey?✓7.98%8%

Traps in the data 7 of 8 handled

  • ✓14.9% of respondents report at least one visit between 11pm and 5am in the last four weeks, against 6% of members in the door-entry records: members with a stake in the overnight hours were more likely to respond. Unweighted, 14.6% say they would be likely to cancel; weighted to the 6%, 11.4%.
  • ✓'Likely to cancel' answers (45.8% of overnight users, 14.6% overall) are stated intentions about a hypothetical closure, asked straight after members were invited to think about it.
  • ✓The mean number of overnight visits among overnight users (7.4) is pulled up by night-shift workers who come almost daily; the median is in the '2 to 3' band.
  • ✓The operations director believes almost nobody uses the gyms overnight. Partly true: 6% of members in the door-entry records (about 2,316 people). But night-shift workers (5.9% of respondents) mostly train between 11pm and 5am (77.8% of those who visited overnight usually arrive then) and are the most likely to say they would cancel.
  • ✓At Wakefield Calder Park (base 11), 54.5% of respondents used the gym overnight and 27.3% say they would be likely to cancel.
  • ✓North West England respondents (base 229) look more likely to cancel than North East England respondents (base 220): 19.2% against 13.6%, a gap that is not statistically significant at 95%.
  • ✓Only 950 of 11,900 invited members completed (about 8%); those who responded differ from those who did not, as the overnight-use skew shows.
  • ✗Only current members were surveyed: people who joined for 24-hour access and have since left, and people who would join only because of it, are missing, so the survey cannot show what 24-hour opening does for recruitment or the brand; and night-shift workers are a small group whose answers carry little weight in the totals even though they depend on the hours.

The analysis

Recommendation

Keep all 14 gyms open 24 hours for now; do not approve a blanket closure or a North West-wide exemption on this evidence. Treat selective closure at genuinely quiet gyms as the next option to evaluate, rather than as a decision already justified. Validate actual site-level overnight demand, including members already inside at 11pm, and compare avoidable savings with lost membership contribution, fee concessions, displacement and implementation costs. If that business case is promising, run a reversible selective-closure pilot with feasible alternatives for affected members and matched 24-hour control sites. Do not designate a site as quiet from these survey percentages alone. The evidence supports caution about blanket closure, but does not establish that permanent 24-hour opening everywhere is financially optimal.

What would change it

Evidence of sustained, low overnight occupancy at particular gyms, meaningful genuinely avoidable costs, accessible alternatives for affected members, and a controlled pilot showing positive net contribution after cancellations, concessions and displaced demand would support selective closure. Blanket closure would require those conditions across all 14 gyms plus evidence that network-wide brand and acquisition effects do not erase the savings. Conversely, negligible avoidable savings, substantial incremental churn, inability to accommodate shift workers, or harmful congestion and acquisition effects would support retaining 24-hour access throughout.

Findings

  1. Operations: overnight use is a minority behaviour, but 'almost nobody' understates its importance. These results do not justify closing all 14 gyms overnight.

    High confidenceBase 950

    Door-entry records identify about 2,316 members, or 6% of the membership, entering overnight in September. In the survey, 142 of 950 respondents reported overnight use: 14.9%. Among those 142, the median falls in the 2–3 visits band, while the mean is 7.4, indicating a smaller group of frequent users. Survey overnight users are substantially overrepresented relative to the administrative benchmark.

  2. Marketing: the survey identifies material cancellation risk, not a reliable forecast of how many members will leave.

    Low confidenceBase 950

    Among 950 respondents, 14.6% were fairly or very likely to cancel. This rises to 45.8% among 142 overnight users, versus 9.2% among 808 non-users. Rebalancing these groups to 6% and 94% gives 0.06 × 45.8% + 0.94 × 9.2% = 11.396%, approximately 11.4%. Applied to 38,600 members, this is approximately 4,400 members expressing cancellation likelihood, not 4,400 predicted cancellations.

  3. The commercial risk extends beyond members currently using the overnight hours. Reduced perceived value and loss of flexibility could affect daytime members too.

    Medium confidenceBase 950

    24-hour opening was a joining reason for 24.0% of all 950 respondents, including 17.8% of 808 recent non-users. Among recent non-users, 9.2% were likely to cancel. Approximately 74 of the 139 respondents likely to cancel were recent non-users. Open texts include demands for lower fees and concerns about crowding immediately before closing and after reopening. These comments explain possible mechanisms but do not establish their population prevalence.

  4. Night-shift workers are a particularly exposed group; any closure plan needs a workable alternative for them rather than assuming everyone can change their training time.

    Medium confidenceBase 56

    Of 56 respondents working shifts that include nights, 80.4% reported overnight use, 62.5% usually arrived overnight and 48.2% were likely to cancel. Sixteen open-text respondents described overnight hours as their only training opportunity because of shift or night work. Nineteen overnight-user text respondents explicitly said they would move to another gym; some also described shift constraints.

  5. Regional manager: do not preserve 24-hour opening throughout the North West solely because its cancellation percentage is higher. The observed difference is not statistically significant, and that does not establish that the regions are equivalent.

    Medium confidenceBase 449

    Likely cancellation was 19.2% in the North West, base 229, versus 13.6% in the North East, base 220: a gap of about 5.6 percentage points. A conventional two-sided independent-proportions test gives z approximately 1.59 and p approximately 0.11. An approximate 95% interval for the difference spans roughly -1.3 to +12.4 percentage points. Overnight-use shares were also similar: 14.0% and 15.0%, respectively.

  6. Assess selective closure by individual gym economics and actual usage, not by regional labels or a survey ranking of the quietest sites.

    Low confidenceBase 950

    Leeds Kirkstall reported 7.5% overnight use and 8.8% likely cancellation, base 80; Bradford Centre 11.5% and 6.6%, base 61; Middlesbrough Centre 9.9% and 8.6%, base 81. These are possible investigation sites, not established closure candidates. Preston Centre illustrates why low use alone is insufficient: 7.7% overnight use but 19.2% likely cancellation, base 78. Wakefield's 54.5% overnight-use result rests on only six users among 11 respondents and should not drive a site decision.

  7. Closure could create costs other than cancellation: demands for price reductions and congestion around the new opening boundaries. Safety concerns should be investigated separately rather than treated as proof that closure is necessary.

    Medium confidenceBase 239

    The 239 open texts include requests for fee reductions, anticipated crowding around 10pm and 5am, and positive reactions based on feeling unsafe in an unstaffed gym. These are reported expectations and perceptions, not measured price sensitivity, capacity effects or evidence of overnight incidents. Some members describe sessions starting before 11pm that could be interrupted by an 11pm closure.

  8. The survey is useful for identifying risks and mechanisms, but its population estimates require substantial caution.

    High confidenceBase 950

    Only 950 of 11,900 invited members completed, a 7.98% completion rate. The unweighted overnight-use share is approximately 2.5 times the 6% administrative benchmark. Its conventional 95% sampling margin is about ±2.27 percentage points, but this does not cover non-response bias. Reweighting on overnight use alone cannot demonstrate that either users' or non-users' cancellation intentions are representative.

Caveats

  • The invitation sample was random, but the completed sample was self-selected. There were 1,031 starts and 950 completes from 11,900 invitations. Low response does not itself quantify bias, but the discrepancy with entry records is a concrete representativeness warning.
  • The 6% benchmark measures September entries, whereas Q1 measures self-reported presence at any time overnight during each respondent's preceding four weeks. These differ in timing and definition: someone entering before 11pm and staying later can meet Q1 without making an overnight entry.
  • The weighted estimate uses the displayed subgroup percentages and assumes the 6% entry benchmark is an appropriate weighting target. It corrects only the overnight-user mix, not selection bias within either group. No database distributions by gym, working pattern or tenure were supplied.
  • The approximately 4,400-member projection is a scenario based on stated intentions, not a forecast of actual losses. Respondents were not given a fee change, alternative-site access, mitigation offer or a selective-closure policy. Do not apply it unchanged to a subset of gyms.
  • The ±2.27-point margin uses 142 of 950 respondents and a conventional normal-approximation calculation without a finite-population correction. It describes nominal sampling uncertainty, not non-response, recall or measurement error.
  • Regional significance was assessed with an unadjusted, two-sided independent-proportions test. Numerous possible subgroup comparisons increase false-positive risk. A non-significant result does not prove there is no commercially meaningful regional difference.
  • Gym-level bases are small, especially Wakefield at 11. The sample's home-gym proportions cannot be judged representative without population membership counts by gym. Home gym is not necessarily the gym used overnight.
  • Only 239 respondents supplied open texts, approximately 25.2% of completes. Themes can overlap and should not be treated as representative prevalence estimates. The switching count requires an explicit move to another gym; cancellation alone is not counted as switching.
  • Some answers warrant a data-quality audit: M034, M305 and M735 are classified as overnight users but deny night-time use in their text. Several respondents mention locations different from their recorded home gym, and M389 describes going after work despite being classified as not in paid work. Cross-gym use and changing circumstances are possible explanations; these are not sufficient grounds for automatic exclusion.
  • Joining reasons are retrospective, multiple-response answers, not a trade-off experiment. Selecting 24-hour access does not establish that a member would leave without it.
  • No site-level avoidable costs, contribution margins, occupancy, competitor availability, incident rates or observed response to changed hours were supplied. The survey cannot establish the financial winner among the three options.

Next steps

  • decisionWithhold blanket closure approval and agree site-level commercial and member-impact criteria before choosing any pilot sites.
  • researchAudit several months of entry records by actual gym, hour and unique member; include exits or another occupancy measure to capture people remaining after 11pm.
  • researchAudit survey linkage and coding, compare respondents with the membership database by gym and tenure, and obtain respondent-level data for weighting and sensitivity analysis.
  • researchBuild a site-level financial model using genuinely avoidable overnight costs and membership contribution, not gross revenue or allocated costs that remain after closure.
  • researchInterview affected overnight users, especially shift workers, and test concrete selective-closure offers with members, including alternative Corefit access, travel requirements and any price change.
  • experimentIf the validated business case supports it, run a reversible selective-closure pilot with comparable 24-hour controls and a pre-specified evaluation covering several monthly renewal cycles.
  • monitorTrack cancellations and their reasons, freezes, transfers, fee concessions, complaints, occupancy near 11pm and 5am, new-member conversion and perceptions of the 24-hour promise.
  • researchReview overnight incidents, emergency support and member safety perceptions independently of the opening-hours decision.

Open-text themes it coded

Explicitly reports no impact on own use or no direct disruption, sometimes alongside another concern 94Expects a lower membership fee or objects to paying the same for fewer hours 52Anticipates crowding or loss of quiet sessions around closing and reopening 31Welcomes closure because unstaffed overnight access feels unsafe or unsettling 42Shift or night work makes overnight hours the only feasible training time 16Explicitly would or probably would switch to another gym 19

The survey it planned

1 screening question and 10 questions, as the model wrote them.

  1. S1

    Corefit is reviewing its overnight opening hours. We want to hear from members with different views, whether or not they use the gyms overnight. This survey takes about 5–6 minutes. Your answers will be linked to your home gym from your membership record so that we can compare results across gyms. Are you currently a Corefit member?

    One answer
    • Yes
    • No
    • Not sure

    Continues if Yes

  2. Q1

    During September 2026, on how many nights did you use any Corefit gym at any time between 11pm and 5am? Include visits that started before 11pm. Treat each 11pm–5am period as one night and count it against the date on which 11pm fell.

    One answer
    • None
    • 1 night
    • 2–3 nights
    • 4–7 nights
    • 8–11 nights
    • 12–19 nights
    • 20–30 nights
    • I used a gym during these hours, but cannot remember how many nights
    • I cannot remember whether I used a gym during these hours
    • I was not a Corefit member during September 2026
  3. Q2

    Which Corefit gym or gyms did you use between 11pm and 5am during September 2026? Please give each gym's name or location. If you cannot remember, write 'Cannot remember'.

    Open text

    Routing Shown only if Q1 is: 1 night; 2–3 nights; 4–7 nights; 8–11 nights; 12–19 nights; 20–30 nights; I used a gym during these hours, but cannot remember how many nights

  4. Q3

    Why did you use Corefit between 11pm and 5am during September 2026? Select all that apply.

    Any that apply
    • To fit around my paid work hours
    • To fit around education or training
    • To fit around caring responsibilities
    • To fit around other commitments
    • I prefer exercising late at night or early in the morning
    • The gym is quieter during these hours
    • To exercise with someone who was available during these hours
    • Another reason
    • Cannot remember — select this option only

    Routing Shown only if Q1 is: 1 night; 2–3 nights; 4–7 nights; 8–11 nights; 12–19 nights; 20–30 nights; I used a gym during these hours, but cannot remember how many nights

  5. Q4

    Thinking back to when you chose your current Corefit membership, which of the following were reasons for choosing Corefit? Select all that apply.

    Any that apply
    • Membership price
    • Rolling monthly membership rather than a longer commitment
    • Convenient location
    • 24-hour opening
    • Equipment or facilities
    • Classes
    • The gym's atmosphere
    • Recommendation from someone else
    • Friends or family were members
    • Another reason
    • None of these — select this option only
    • Cannot remember — select this option only
  6. Q5

    Imagine that all 14 Corefit gyms closed every night between 11pm and 5am, but remained open between 5am and 11pm. Membership prices and all other membership terms would stay the same. How much would this affect your ability to exercise at a Corefit gym when you want to?

    One answer
    • Not at all
    • A little
    • Moderately
    • A lot
    • I would no longer be able to exercise at Corefit
    • Not sure
  7. Q6

    Under that same situation—all Corefit gyms closed between 11pm and 5am, with membership prices and all other terms unchanged—what would you be most likely to do?

    One answer
    • Definitely keep my Corefit membership
    • Probably keep my Corefit membership
    • Probably cancel my Corefit membership
    • Definitely cancel my Corefit membership
    • Not sure
  8. Q7

    Now imagine a different situation: only your home gym—the Corefit gym listed as your main gym on your membership—closed every night between 11pm and 5am. All other Corefit gyms would remain open 24 hours. Membership prices and your existing rules about access to other gyms would stay the same. How much would this affect your ability to exercise at a Corefit gym when you want to?

    One answer
    • Not at all
    • A little
    • Moderately
    • A lot
    • I would no longer be able to exercise at Corefit
    • Not sure
  9. Q8

    Under that second situation—only your home gym closed between 11pm and 5am, with all other Corefit gyms open 24 hours and membership prices and terms unchanged—what would you be most likely to do?

    One answer
    • Definitely keep my Corefit membership
    • Probably keep my Corefit membership
    • Probably cancel my Corefit membership
    • Definitely cancel my Corefit membership
    • Not sure
  10. Q9

    Which of these currently apply to you? Select all that apply.

    Any that apply
    • I do paid work between 11pm and 5am
    • I do paid work with changing or rotating hours
    • I do paid work with regular hours outside 11pm–5am
    • I am in education or training
    • I have regular unpaid caring responsibilities
    • None of these — select this option only
    • Prefer not to say — select this option only
  11. Q10

    What is your age?

    One answer
    • Under 18
    • 18–24
    • 25–34
    • 35–44
    • 45–54
    • 55–64
    • 65 or over
    • Prefer not to say
Sample plan

Invite a stratified random sample of current members from the membership database, using home gym as the 14 sampling strata. Target 1,000 eligible completes: 72 from each of the six largest home-gym populations and 71 from each of the remaining eight. Randomly order members within each stratum and release invitations in batches; do not replace nonresponders with volunteers or staff-selected members. Invitation volume depends on response rate: at an illustrative 20% eligible completion rate, approximately 5,000 invitations are needed. Monitor each gym and send consistent reminders and additional random batches where necessary. Use a neutral invitation about opening hours, not an invitation framed around overnight users or support for closure. Retain a pseudonymous member identifier and home gym, and derive region from gym location. Weight chain-wide estimates to the current membership distribution across gyms; assess differential response by September overnight-entry status where available. Randomise the substantive options in Q3 and Q4, keeping 'Another reason' and exclusive answers at the bottom; enforce exclusive answers. Pilot on mobile and desktop to verify comprehension and completion within six minutes. PRECISION LIMITATION: 1,000 completes cannot generally deliver ±8 percentage points at 95% confidence for every gym. With 71–72 respondents per gym, the worst-case margin is about ±11.6 percentage points before finite-population correction or weighting effects. Without finite-population correction, ±8 points requires 151 completes per gym, or 2,114 total. Obtain actual member counts by gym and calculate finite-population-adjusted requirements using n = ceiling[N × 1.96² × 0.25 / (0.08² × (N − 1) + 1.96² × 0.25)]. Any weighting design effect can increase the requirement. Seek agreement either to increase completes to the calculated total or to accept wider gym-level intervals; do not claim that 1,000 meets the requested precision. The ±8-point target applies to whole-gym proportions, not overnight-user subgroups, and does not cover nonresponse bias.