Member Retention
Paid community survey — three tenure-timed surveys, nine questions, reference card
Survey timing determines what problem the data describes. A 30-day member and a 10-month member are structurally different respondents with structurally different problems; averaging their answers into one satisfaction score produces data about neither. This reference card covers the three tenure-timed surveys for paid community operators — the 30-day activation survey, the 90-day contribution-gap survey, and the 10-month renewal-risk survey — with three specific questions per survey, a timing summary table, the question-type comparison grid (how each question type is framed differently at each tenure stage), and the wrong survey patterns that produce data operators cannot use.
TL;DR
Three surveys, three questions each, nine total. 30-day activation survey: sent at day 30 to every member; diagnoses expectation-gap and activation-lag before they become month-three churn. 90-day contribution-gap survey: sent at day 90; identifies the passive-observer cohort (logged in, never posted) before they reach the months-5–7 programming void. 10-month renewal-risk survey: sent at month 10; renewal-intent score 1–10 plus open-ended gap question plus peer-relationship check — the strongest predictor of year-one renewal is a yes to “can you name a specific peer connection this community produced?” Run all three as tenure-triggered events (every member gets each survey when they hit the milestone), not as quarterly broadcast campaigns. The annual broadcast survey is structurally unsuited to this diagnostic task — it averages together populations with opposite problems.
Survey timing summary table
Each survey is designed to diagnose the failure mode specific to that tenure window. Sending the wrong survey at the wrong time produces data about a different problem than the one the operator needs to solve.
| Survey | When to send | Trigger | Diagnoses | Intervention window | Response channel |
|---|---|---|---|---|---|
| 30-day activation survey | Day 30 from join date | Tenure milestone, every new member | Expectation-gap + activation-lag before month-three churn | 60 days before months-2–3 peak exit | Slack DM (preferred) or email |
| 90-day contribution-gap survey | Day 90 from join date | Tenure milestone, every surviving member | Passive-observer cohort before months-5–7 programming void | 90 days before programming-void exit window | Slack DM (preferred) or email |
| 10-month renewal-risk survey | Month 10 from join date | Tenure milestone, year-one cohort | Relationship-thin non-renewal before the annual billing date | 60–90 days before renewal decision forms at month 12 | Email (preferred for low-frequency users) + Slack DM follow-up |
Why timing is the structural variable. A survey is retrospective by design — it describes what has already happened. The question is whether it arrives early enough that operator action can change the trajectory. The 30-day survey arrives when a month-three exit is still 60 days away. The 10-month survey arrives when an annual non-renewal is 60–90 days away. A survey sent at month 12 — or annually, averaging all tenure stages — arrives after the renewal decision has already been made. The data describes a decision, not a problem the operator can still fix.
The 30-day activation survey
The 30-day survey diagnoses two distinct failure modes that both show up as month-one-to-two churn but have different root causes and different fixes. Expectation-gap exits (days 7–14) occur when members joined expecting something the community does not deliver. Activation-lag exits (days 21–28) occur when members intended to participate but could not find a specific entry point. Without a structured survey at day 30, operators cannot distinguish which failure mode is driving their month-one numbers. For the full churn diagnostic across all four tenure windows, see the paid community member churn by tenure reference card.
Activation survey — three questions
-
Q1
Expectation question “What made you decide to join?” An open-ended question, not a multiple-choice rating. The answer tells you what the member expected before joining — not whether they are satisfied after joining. Open-ended answers reveal the expectation that was or was not met: “I joined to meet other founders who are earlier-stage than the YC forums” or “I was hoping for more practical tooling advice rather than strategy discussions.” Those are diagnostic signals you cannot extract from a satisfaction score. Code the answers into two categories: expectation-aligned (the community delivered what the member anticipated) and expectation-misaligned (the member expected something different). Month-one churn concentrated in the expectation-misaligned group is a value-proposition problem; churn distributed across both groups is an activation-sequence problem.
-
Q2
Contribution question “Have you posted or contributed to a discussion in the past four weeks?” A binary yes/no question, not a frequency or quality rating. The answer is the most reliable 30-day predictor of 3-month renewal: members who answer “yes” renew at 2–3× the rate of members who answer “no.” The follow-up for no-responders is a different action than the follow-up for yes-responders who nevertheless scored low on Q3. Members who answer “no” at day 30 are the activation-lag cohort; the intervention is an operator DM offering a specific, lower-barrier contribution entry point. Do not ask “how often do you contribute?” — frequency-rating questions at 30 days introduce social-desirability bias that makes the data noisier without making it more useful.
-
Q3
Friction question “What’s the one thing that would make this community more useful to you in the next 30 days?” A forward-looking friction-naming question. The constraint “one thing” forces specificity: members name the most pressing gap, not a wishlist. Answers cluster into three categories: content friction (the community doesn’t discuss the specific topics I joined for), structural friction (I don’t know where to find the conversations I want), and relationship friction (I haven’t connected with anyone who works on my specific problem). Each category has a different operator fix. Content friction is a programming problem. Structural friction is a channel-architecture or welcome-sequence problem. Relationship friction is a peer-introduction problem — and relationship friction at day 30 predicts the relationship-thin year-one non-renewal before it forms. A member who names relationship friction at 30 days and receives no peer introduction by day 90 is on a predictable path to non-renewal.
What this survey does NOT tell you
Whether members will renew at month three. The 30-day survey identifies the failure modes and provides 60 days of intervention time — but renewal probability at month three is determined by what the operator does between day 30 and day 60, not by the survey data itself. A high friction-friction response rate on Q3 is not a prediction of churn; it is a list of operator actions. The survey also does not distinguish between members who cancel in weeks one through two (expectations-misalignment exits) and members who cancel in weeks three through four (activation-lag exits) — that distinction requires cross-referencing Q1 and Q2 answers with the actual cancellation date from billing records.
The 90-day contribution-gap survey
The 90-day survey diagnoses the passive-observer cohort: members who logged in regularly, read threads, and may have attended events, but never posted or contributed. This cohort is invisible in most community health metrics because they show up as “active” in login-based engagement reports. But their non-contribution predicts months-4–6 exit at substantially higher rates than contribution-active members. The 90-day survey is the mechanism that surfaces them before the programming void — the stretch from months four through six where orientation content is exhausted and operators have no specific programming intervention in place — produces churn that looks unexplained. For the activation rate benchmarks that should accompany this cohort diagnostic, see the companion reference card.
Contribution-gap survey — three questions
-
Q1
Contribution question “What’s the main way you’ve engaged with the community in the past three months?” A multiple-choice question with four options: (a) I’ve posted threads or replies, (b) I’ve attended live events or sessions, (c) I’ve mostly read or lurked, (d) I’ve messaged specific members directly. The question identifies the engagement modality, not the quality or frequency. Members who select (c) are the passive-observer cohort. Members who select (b) but not (a) are event-consumers who have not contributed to asynchronous discussions — a related but distinct risk group. Members who select (d) are peer-relational engagers whose renewal is better predicted by Q3. The multiple-choice format at 90 days is preferable to open-ended here because you want to categorize the cohort quickly, not to diagnose root cause — that comes from Q2 for the (c) group and the cross-reference with Q3 for the (d) group.
-
Q2
Friction question “Has there been a topic you wanted to raise in the community but didn’t? If so, what stopped you?” A two-part question: the first part is binary (yes/no), the second part is open-ended. Members who answered (c) on Q1 and answer “yes” on Q2 are the highest-priority intervention group: they have something to contribute and a specific reason they didn’t. The open-ended answers to “what stopped you?” cluster into: social risk (I wasn’t sure if my question was relevant or too basic), structural ambiguity (I didn’t know which channel to post in), and bandwidth friction (I kept meaning to post but ran out of time). Social-risk friction is addressed by the operator lowering the bar explicitly in a DM (“there are no too-basic questions in this community”). Structural ambiguity is addressed by giving the member a specific channel and thread. Bandwidth friction is addressed by offering an async contribution that takes less than 60 seconds (a reaction, a one-sentence reply to a named question). Members who answer (c) and “no” to Q2 — lurking by preference rather than by friction — are a different cohort with lower intervention urgency.
-
Q3
Peer question “Which member or conversation has been most useful to you in the past three months?” An open-ended question that functions as a peer-relationship diagnostic. Members who can name a specific member (not a type of member, not a channel, but a named individual) are beginning to develop the peer relationship that predicts year-one renewal. Members who cannot name anyone, or who name a channel or content type rather than a specific member, are relationship-thin at 90 days — and relationship-thin members at 90 days are the same population who will appear in the 10-month renewal-risk survey as non-renewal candidates. The 90-day peer-relationship diagnostic is valuable because it identifies the at-risk year-one population 10 months before their billing date, when a facilitated peer introduction can still produce a real relationship. An operator who introduces two members at month three with 10 months of shared community context ahead of them is doing fundamentally different work than an operator who makes the same introduction at month 11 with two weeks before the billing date.
What this survey does NOT tell you
Why a member who was fully active in months one through two went passive in months two through three. The 90-day survey identifies the passive-observer cohort but does not diagnose the transition point. That diagnosis requires cross-referencing the survey data with the member’s Slack activity log: did they post actively in week two and then stop? Did they attend the month-two event but not the month-three event? The transition point tells you which specific operator action (or absence of action) triggered the drift. The survey names the problem; the activity log finds the cause.
The 10-month renewal-risk survey
The 10-month survey is the only tenure-timed survey where individual responses are immediately actionable without a sample threshold. A single member with a renewal-intent score of 3/10 is an intervention target regardless of what the rest of the cohort says. The survey is designed to identify at-risk members 60–90 days before their annual billing date, when operator action can still produce a peer relationship that changes the renewal calculus. A discount offer or re-engagement campaign at month 12 reaches members who have already evaluated a year of experience and made a decision — the intervention window is closed. For the full programming calendar that places this survey in the year-one operator action sequence, see the companion reference card.
Renewal-risk survey — three questions
-
Q1
Renewal-intent question “On a scale of 1–10, how likely are you to renew your membership next month?” A 1–10 intent score, not a satisfaction score. The intent-vs-satisfaction framing is deliberate: satisfaction scores at month 10 predict satisfaction, not renewal. A member can be satisfied with the community they used in months one through four and still not renew at month 12 because the membership has become passive. The intent score produces a triage list: members scoring 7–10 are renewals (no immediate action needed beyond a thank-you DM from the operator); members scoring 4–6 are uncertain (the gap identified in Q2 is solvable before the billing date); members scoring 1–3 are at-risk and require a personal operator call or DM within 48 hours of the survey response. The 1–10 format is preferable to a binary yes/no because it reveals the at-risk gradient: a 4 and a 2 require different urgency, and treating them identically wastes operator time on the 4 and loses the 2.
-
Q2
Gap question “What’s the main thing that would make you more likely to renew?” An open-ended question for members scoring below 8 on Q1 (members scoring 8–10 can be skipped on this question or offered it optionally). The answers reveal the specific gap between what the member experienced and what would make another year feel worth the price. Answers cluster into three categories: content gap (the community stopped discussing the topics I joined for), relationship gap (I haven’t connected with anyone I’d call a peer or collaborator), and ROI gap (I can’t name a specific outcome the community produced for my work). Each has a different operator response: content gap requires programming changes before the billing date; relationship gap requires an immediate peer introduction in months 10–11; ROI gap requires an operator-facilitated articulation of value (“let me remind you what changed in your work this year”). The relationship gap is both the most common answer and the most fixable within 60 days — one well-matched peer introduction produces a relationship; a discount offer does not.
-
Q3
Peer question “Have you made a specific connection through this community that you’d describe as a peer or colleague?” A binary yes/no question. This is the single strongest predictor of year-one renewal in paid communities. Members who answer “yes” renew at 70–80%; members who answer “no” renew at 40–50% regardless of satisfaction score, engagement history, or content consumption. A member who attended every event and read every thread but cannot name a peer is in the relationship-thin category. The operator action for a “no” at month 10 is an immediate facilitated introduction to one specific member — not a cohort invitation, not a networking event, but a direct operator DM matching two specific members based on a shared problem the operator observed in 10 months of watching both of them contribute. A “no” at month 10 that receives a well-matched introduction by month 11 has meaningful probability of converting to a “yes” at renewal. A “no” at month 10 that receives nothing converts at the baseline 40–50% rate.
What this survey does NOT tell you
Whether a relationship-thin member will respond to an operator-facilitated introduction in time to affect the renewal decision. The intervention window is 60–90 days (months 10–11). A peer introduction that produces a real conversation within three weeks has enough time to compound into a relationship before the billing date. A peer introduction made two weeks before billing does not — the member evaluates a year of relationship-thin experience, not a two-week acquaintance. The survey identifies the at-risk population and the problem; the timing of the operator response determines whether the intervention lands in the window or after it.
Question-type comparison across all three surveys
Each survey uses the same three question types — expectation, contribution, and peer — framed differently at each tenure window because the framing that produces useful data changes as the member’s tenure arc changes. The expectation question at 30 days names the gap; the same question asked at 10 months names what the member gave up on. The peer question at 30 days is absent (no real peer relationships form in the first month); at 90 days it identifies early-stage relationship formation; at 10 months it is the primary renewal predictor.
| Question type | 30-day framing | 90-day framing | 10-month framing | What changes across windows |
|---|---|---|---|---|
| Expectation question | “What made you decide to join?” — identifies pre-join expectation before it’s revised by 30 days of experience | Omitted at 90 days; replaced by the contribution-type question which is more diagnostic at this window | Embedded in Q2 (“what would make you more likely to renew?”) — retrospective gap naming after 10 months | Moves from prospective expectation (what did you expect) to retrospective gap (what didn’t happen). The 10-month version names what the member gave up on, not what they hoped for. |
| Contribution question | “Have you posted or contributed in the past four weeks?” — binary yes/no, strongest 3-month renewal predictor at this window | “What’s the main way you’ve engaged?” — multiple choice, identifies engagement modality (contributor vs. consumer vs. peer-relational) | Implicit in Q1 (renewal-intent score) — members who contributed actively renew at higher rates; the score captures the outcome without restating the activity | Moves from binary activation (did you post?) to modality identification (how did you engage?) to outcome inference (did your engagement produce the value that drives renewal?). Framing evolves with the diagnostic question. |
| Peer question | Absent — no meaningful peer relationship forms in 30 days; the question would produce noise, not signal | “Which member or conversation has been most useful to you?” — identifies emerging peer-relational engagement; names at-risk year-one population 10 months early | “Have you made a specific connection you’d describe as a peer or colleague?” — binary, the single strongest predictor of year-one renewal | Moves from absent (too early) to qualitative-naming (who has been useful?) to binary-predictive (do you have a named peer?). The peer question earns its predictive weight only after 10 months of experience — at 30 days it would be a leading question with no answer to give. |
Wrong survey patterns
Most paid community operators run one of three wrong survey patterns that produce data they cannot use. The problem is structural: each wrong pattern optimises for a different dimension (reach, familiarity, or retrospective precision) that is not the diagnostic dimension the operator needs at any specific tenure stage.
| Wrong pattern | What it optimises for | Why the data is unusable | What to do instead |
|---|---|---|---|
| Annual broadcast survey | Reach — sent to all members once per year; high absolute response count | Aggregates responses from members at wildly different tenure stages. A 30-day member and a 10-month member have structurally different problems; averaging their answers produces a score that describes neither. The same aggregate satisfaction score can describe a community with strong year-one retention and dismal month-one activation and a community with the opposite profile — they look identical in an annual survey and require opposite interventions. | Replace the annual survey with three tenure-triggered surveys sent to each member individually when they hit 30 days, 90 days, and 10 months. Same total survey volume; dramatically higher diagnostic value. |
| NPS-only survey | Familiarity — NPS is a format operators and members both recognize; easy to report | NPS measures likelihood to recommend, which correlates with satisfaction but does not identify which tenure-stage failure mode is driving dissatisfaction. A community with identical NPS scores can have month-one churn driven by expectation-gap or month-three churn driven by contribution-gap — the NPS score is the same; the operator intervention is completely different. NPS also produces the aggregation problem: a 30-day “detractor” and a 10-month “detractor” are detractors for different reasons that require different fixes. | Use NPS as a quarterly aggregate health indicator, not as the primary diagnostic instrument. Pair it with tenure-timed surveys that name the specific failure mode the NPS score cannot. |
| Cancellation-only exit survey | Retrospective precision — asks cancelling members why they left; data is specific and named | Describes decisions already made. The member who fills out a cancellation survey has already cancelled; the operator is collecting a post-mortem, not a prevention signal. Cancellation surveys are useful for identifying patterns across a trailing 6-month cancellation cohort, but they produce no intervention window for the individual members who cancelled. Additionally, only members motivated enough to respond fill out exit surveys — typically the most recently engaged members, which introduces selection bias that makes the data less representative of the silent-exit majority. | Use the cancellation survey for trailing-cohort pattern recognition (what percentage of month-one cancellations named expectation-gap?), not as the primary retention tool. The primary prevention tool is the 30-day survey, which arrives 60 days before the month-three churn peak and names the same failure modes before they produce exits. |
The aggregation problem in one sentence. Any survey that mixes respondents at different tenure stages is measuring the average of different problems — and the intervention for an average of different problems is an average of different interventions, which is no intervention at all.
Frequently asked questions
What is the difference between a tenure-timed survey and a standard satisfaction survey?
A standard satisfaction survey asks members how they feel about the community in general — producing an aggregate score averaged across all tenure stages. A tenure-timed survey is sent to a specific cohort at a specific tenure window (30 days, 90 days, 10 months) and asks questions that are only meaningful at that stage. The 30-day survey asks about activation and expectation-gap — questions that are useless at 10 months because a 10-month member has revised their expectations a dozen times. The 10-month survey asks about peer relationships and renewal intent — data that 30-day members cannot yet provide. Averaging a 30-day member and a 10-month member into the same satisfaction aggregate produces a score that describes neither. Tenure-timed surveys fix this by treating each tenure cohort as a separate research question with separate diagnostic needs.
How many responses do I need before survey data is actionable?
Each tenure cohort needs 10–15 responses for a pattern to be visible and 25–30 responses before the data is reliable enough to drive an operational change. For the 30-day activation survey (sent to every new member), at 5–10 new members per week you reach a usable sample within 3–6 weeks. For the 90-day contribution-gap survey (the surviving fraction of month-one joiners), plan for 6–8 weeks of collection before acting on aggregate patterns. For the 10-month renewal-risk survey, the cohort is smaller — but this survey is useful even with a small number of responses because individual renewal-intent scores on Q1 are actionable at the single-member level. A member scoring 3/10 on renewal intent is an intervention target immediately, regardless of what the rest of the cohort says. The renewal-risk survey is the only case where one response is enough to act.
Should tenure-timed surveys be sent via email or Slack DM?
Slack DM is the higher-response channel for the 30-day and 90-day surveys because the member is already in the workspace and responding to a DM requires less context-switching than clicking through from an email to an external form. For the 10-month renewal-risk survey, email is preferable: members at month 10 often have lower Slack login frequency than they did in months one and two, and a Slack DM to a low-frequency user is more likely to be missed. A reliable sequencing: send the 30-day and 90-day surveys via Slack DM with a single-link form, and send the 10-month survey via email with a Slack DM follow-up 48 hours later for non-responders. Keep each survey to three questions regardless of channel — response rates drop sharply past three questions for unsolicited surveys in a community context.
How do the three tenure-timed surveys relate to each other as a diagnostic system?
Each survey diagnoses the failure mode specific to that tenure window and provides an intervention window before the associated churn event. The 30-day survey diagnoses activation failure 60 days before months-2–3 churn. The 90-day survey diagnoses the passive-observer cohort 90 days before the months-5–7 programming-void exit. The 10-month survey diagnoses relationship-thin non-renewal 60–90 days before the annual billing date. Together they form a sequential early-warning system: each survey’s peer question (Q3 at 90 days, Q3 at 10 months) also tracks whether the peer relationship problem identified early in the member’s tenure arc has been addressed — a member who answers “no one specific” on the 90-day peer question and receives an operator introduction should answer “yes” on the 10-month peer question. If they don’t, the introduction didn’t take, and the operator has 60 days to try again. For the full churn diagnostic context, see the paid community member churn by tenure reference card.
How often should I run each tenure-timed survey?
Each tenure-timed survey should be sent to every member when they hit that tenure milestone — not on a quarterly or annual broadcast schedule. The 30-day survey goes to every new member at day 30 from their join date, as a recurring trigger tied to join date. The 90-day survey goes to every member at day 90, as the same recurring trigger. The 10-month survey goes to every member at month 10 from their join date. This means all three surveys run on rolling schedules: a community operating for 18 months has a steady flow of surveys going out weekly to members who hit the relevant tenure milestone that week, rather than periodic mass surveys that mix tenure stages. The mass-survey format — “we are doing our quarterly survey, please fill this out” — produces the aggregation problem that tenure timing is designed to avoid. Running surveys as tenure-triggered events also improves response rates: a survey that arrives at the right moment in a member’s tenure arc feels relevant, not generic.