Paid community annual review: the six-layer framework
Most paid community operators who want to review their year look at two numbers: MRR at December 31 versus MRR at January 1, and total membership at year-end versus year-start. If both numbers went up, the year was good. If either went down, the year was not. This is the wrong review. Not because MRR and membership count are the wrong metrics — they are the right outcomes to optimize for — but because they are outcome metrics with no diagnostic content. They tell you what happened. They cannot tell you why, and they give you no specific basis for deciding what to change in the coming year.
An operator who ends the year with MRR up 40% has had a good year. But without knowing whether that growth came from activation improvements, lower churn, higher new-member volume, or a pricing change, they cannot tell which of those drivers to reinforce next year. An operator whose MRR is flat cannot tell from that number alone whether the problem is that fewer new members are activating, that more activated members are cancelling at month three, or that acquisition has slowed and the existing membership base is holding fine. Each diagnosis implies a different intervention, and a flat MRR number cannot distinguish between them.
A complete paid community annual review has six layers, and the sequence in which you run them matters as much as the layers themselves. The six layers are: cohort retention rate, activation rate trend, content engagement by channel, moderation load, member NPS trend, and economics. You run them in that order because each layer provides context that makes the next layer interpretable. Running economics first — which is what most operators do, because it is the most concrete number — produces conclusions that cannot be acted on, because the economics are downstream of the operational and community-design decisions that shape them. This guide covers each layer: what to measure, how to produce the number, and what each pattern means for the decisions that follow.
1. Why MRR and membership count are the wrong starting point
MRR and membership count are summaries. They collapse everything that happened across 12 months — every onboarding decision, every content investment, every churn event, every new join — into a single pair of numbers. That collapse is useful for communicating the community’s trajectory to investors, partners, or a board. It is not useful for understanding what drove that trajectory, and therefore not useful for planning the next year.
Consider two communities that both ended the year with 500 members. Community A started the year with 300 members, acquired 400 new members across the year, and churned 200. Community B started with 450 members, acquired 150 new members, and churned 100. Both communities ended at 500. The review conversation that starts with “we hit 500 members” looks identical for both. But the strategic situation is completely different. Community A has a high acquisition rate and a high churn rate — it is a leaky bucket that is growing despite the leak. Community B has a low acquisition rate but high retention — it is holding its members well but not finding new ones. The interventions are opposite: Community A needs to fix onboarding and early retention; Community B needs to fix top-of-funnel or referral mechanics. A review that starts and ends with total membership count will not surface this distinction.
The same problem applies to MRR. Two communities with identical end-of-year MRR can have completely different churn rates if one has a higher-priced plan mix or a larger average cohort size. MRR is an aggregate that masks the distribution of outcomes across member cohorts and plan types. The six-layer review is designed to surface that distribution before reaching the economic summary, so the economics can be interpreted in context.
2. The six review layers and why the sequence matters
The correct sequence is: retention rate by cohort → activation rate trend → content engagement by channel → moderation load → NPS trend → economics. Each layer is a diagnostic tool that frames the layer after it.
Cohort retention rate comes first because it is the most granular window into what is happening at different stages of the member lifecycle. It distinguishes between churn that happens at month two (an onboarding problem) and churn that happens at month six (a programming or value-delivery problem at an intermediate tenure stage). Until you know at which tenure point your members are leaving, you cannot correctly frame the activation rate trend in layer two.
Activation rate trend comes second because activation is the leading indicator for all subsequent retention. A member who activates in their first 30 days — who posts in public channels, attends a live event, and has at least one direct interaction with the operator or a peer member — is substantially more likely to retain at months two, three, and six than a member who did not. The activation trend tells you whether your improvements to onboarding are working, and it contextualises the churn patterns in layer one. High month-two churn in the retention table combined with a flat or declining activation rate is a strong signal that the onboarding sequence itself is the problem, not the community’s content or programming at later tenure stages.
Content engagement by channel comes third because it identifies where the community’s investment of programming, moderation, and curation time is producing returns and where it is not. A channel that the operator posts into weekly but that generates no member-to-member discussion is consuming operator bandwidth without producing community value. This layer frames the moderation load layer that follows it: a community with many dead channels often has a moderation load problem driven by the structural complexity of the workspace rather than by norm violations.
Moderation load comes fourth because it is a leading indicator for community health at scale. Rising incidents per 100 active members, even in a growing community, signals norm degradation — new members are not being adequately socialized into the community’s standards, or the standards themselves are not being enforced consistently. The paid community audit framework covers moderation load in detail; for the annual review, the key question is the trend: is the moderation load per 100 active members higher at year-end than at year-start? If yes, scale is making the community harder to manage, not easier.
NPS trend comes fifth because it is the perception layer. Member NPS — “how likely are you to recommend this community to a peer?” — tells you whether members’ subjective experience of the community is moving in the same direction as the operational metrics in layers one through four. A community with improving retention and activation but declining NPS has a perception gap: members are staying and participating but they are not enthusiastic about the community in a way that generates referrals. A community with declining retention but improving NPS has the opposite gap: members like the community but they are leaving for reasons that have nothing to do with satisfaction (budget, life change, competing platform). These two diagnoses require different interventions, and neither can be identified without both the operational data and the NPS data.
Economics comes last — LTV by cohort, CAC trend, MRR versus target — because those numbers can now be interpreted. If cohort retention is strong but LTV is lower than expected, the issue may be plan-mix (too many members on Starter, not enough upgrading to Pro). If CAC is rising but activation rate is stable, the issue is in acquisition channels, not in what happens to members after they join. The economics layer closes the review by connecting the operational findings to the financial outcomes the operator needs to communicate to stakeholders.
3. Layer 1: cohort retention rate
Produce a table with join month as rows (January through December of the review year) and months-since-join as columns (month 0 through month 11). Each cell is the percentage of that cohort who were still active at that tenure point. A member is active if they made at least one substantive public contribution in the preceding 30 days — a message in a public channel, attendance at a live event, or a reaction count above a minimum threshold you hold constant across all cohorts.
The table you are looking for is not a single number. It is a set of curves — one per cohort — and the diagnostic value is in comparing those curves. The four patterns that matter most:
Consistent downward slope across all cohorts, steepest at months two and three. This is the classic onboarding problem. Every cohort is losing the same fraction of members at the same tenure stage. The intervention is in the first 30 days: the Day 0, Day 3, and Day 7 touchpoints, the quality of the new-member welcome sequence, and whether the activation threshold is achievable within the first month for a new member who arrives with no prior context about the community.
Specific cohorts with steeper drops at months five through eight, adjacent cohorts holding. This is a programming problem at a specific calendar period. The members of the cohort that shows steep mid-tenure drop were at months five through eight during a specific set of calendar months, and something about what the community offered — or failed to offer — during those calendar months drove elevated churn. Review what programming, content, or events ran during that period versus adjacent periods where the adjacent cohorts held better.
All cohorts holding flat through month six, then declining at months eight through twelve. This is a long-term value problem. The community is delivering enough in the first six months that members stay, but it is not delivering enough at the twelve-month horizon to justify renewal. The most common driver is content depth: early members benefited from foundational content and initial community building; later in their tenure, the content is covering ground they already know and the community’s discussions are repeating themes they have already seen addressed. The intervention is programming designed for members who have been in the community for six months or more: advanced tracks, peer accountability cohorts, exclusive senior-member channels.
Newer cohorts retaining better than older cohorts at equivalent tenure points. This is a positive signal: onboarding or programming improvements made during the year are producing better retention in the cohorts that experienced them. Identify the specific change that correlates with the improvement and make it the foundation for the next year’s onboarding investment.
The paid community metrics dashboard guide covers how to structure the data export from Slack and your payment processor to produce this table without custom tooling. For communities using Stripe, the cohort table can be approximated from subscription created-at dates, cancellation-at dates, and message timestamps from the Slack export.
4. Layer 2: activation rate trend
For each calendar month in the review year, calculate: (members who joined that month and crossed the activation threshold within 30 days) ÷ (total members who joined that month). Plot the 12 monthly rates in chronological order.
Activation threshold must be consistent across all months. A common definition for paid communities at the 200–2,000 member range is two or more public messages in any non-announcement channel plus attendance at one live event, all within the first 30 days. This threshold is achievable by a new member who joins with genuine intent and encounters an onboarding sequence designed to prompt both actions. It is not achievable by passive readers or by members who join during periods of low community activity (holiday breaks, for example — which is one reason to check whether December and January cohorts show systematically lower activation rates before concluding there is an onboarding problem).
What a healthy activation rate trend looks like: stable or improving month-over-month, with the improvement correlated in time with specific onboarding changes. If the community made no changes to the Day 0 DM or the welcome sequence in the review year, the activation rate trend should be approximately flat across cohorts acquired through the same channels. Meaningful month-to-month variance in activation rates — where the range spans more than 15 percentage points across the year — typically reflects either acquisition channel mix changes (cohorts from higher-intent channels activate at higher rates than cohorts from lower-intent channels) or seasonal effects (summer and December cohorts with different join-intent profiles from peak months).
The activation rate trend frames everything in the retention cohort table. If you see high month-two churn in the retention table and the activation rate trend is flat or declining, the intervention is in the onboarding sequence: the community is not giving new members the specific actions and social connections that produce activation, and that gap is expressing itself as month-two churn when the members who did not activate quietly stop logging in. If you see high month-two churn but the activation rate is stable or improving, the churn is not driven by onboarding failure — look at whether the members churning at month two were activated members who found the community’s ongoing content insufficient, or unactivated members from a specific acquisition channel with misaligned expectations.
5. Layer 3: content engagement by channel
For each channel in the workspace, measure two things: operator posts per month averaged across the year, and member-to-member replies per operator post averaged across the year. The ratio of member replies to operator posts is the channel’s engagement multiplier. A channel with an engagement multiplier above 3 — every operator post generates at least three member responses — is a channel where members are finding the content worth responding to. A channel with a multiplier below 1 is a channel where the operator is posting and members are reading without engaging.
The annual review question is not which channels have the lowest engagement multiplier in any given week — it is which channels have consistently low multipliers across the whole year, despite operator investment. A channel the operator posts into weekly but that consistently generates fewer than two member replies per post is a candidate for reduction, restructure, or elimination. In most paid Slack communities at the 200–1,000 member range, the 80/20 rule applies strongly to channel engagement: 20% of channels produce 80% of member-to-member discussion. Identifying which channels are in that 20% and doubling down on them, while reducing investment in the low-multiplier 80%, produces more community value from the same operator bandwidth.
The content engagement review also surfaces which topics the community most values by proxy. If the highest-multiplier channel covers pricing strategy, that signal tells you the community values that topic enough to discuss it actively. If the lowest-multiplier channels cover process and tooling topics, that signal tells you members read but do not engage with that content — they find it useful for reference but do not have responses to contribute. That distinction — reference content versus discussion-generative content — should inform the content calendar for the coming year.
The paid community survey guide covers how to supplement the engagement data with a member survey that asks directly which channels members find most valuable. Combining the engagement multiplier data with the survey responses produces a clearer picture than either source alone: some channels have high engagement multipliers because a small number of highly active members post in them frequently, but a member survey may reveal that the broader membership does not find those channels relevant to their reason for joining. Those channels are candidates for restructure to broader appeal rather than simple continuation or elimination.
6. Layer 4: moderation load
Moderation load is incidents per 100 active members per month, averaged across the review year. An incident is any action the operator or a designated moderator had to take in response to a norm violation: a message deleted, a member warned, a member removed, a thread locked, or a dispute between members that required moderator intervention. Track these as they happen, not retroactively — a post-year attempt to reconstruct moderation incidents from memory or Slack message history will undercount.
The benchmark for well-functioning paid communities at the 200–1,000 member range is below 0.5 incidents per 100 active members per month. Above 1 incident per 100 active members per month consistently across the year suggests a norm health problem. Rising incidents per 100 active members over the course of the year — even from a low starting point — is a more important signal than any absolute level: a community that goes from 0.2 to 0.6 incidents per 100 active members across the year is heading in the wrong direction regardless of the absolute level being below 1.
Rising moderation load at scale has two common causes. The first is onboarding norm failure: new members are joining without adequate socialisation into the community’s norms, and they are exhibiting behaviours that would be self-correcting in a community where new-member socialisation is built into the onboarding flow. If new-member incidents are higher per capita than long-tenure member incidents, this is the likely cause. The intervention is in the onboarding sequence: the Day 0 DM should explicitly reference community norms, and the Day 3 nudge should include a norm-relevant prompt. The second cause is community norm drift at scale: as communities grow past 500 members, the informal social enforcement mechanisms that worked at smaller scale — long-tenured members setting the tone by example, the operator being visible enough that members calibrate to their presence — weaken. The intervention is formalised community standards documents, explicit moderator roles (not just the operator), and a clearer escalation path for members who observe norm violations.
7. Layer 5: NPS trend
Member NPS for the annual review requires at least two survey points: one early in the review year and one late. A single end-of-year NPS tells you the current level but gives you no directional signal. The most useful configuration is a quarterly survey cadence — four NPS data points across the year — which gives you enough resolution to see whether NPS is moving and whether it correlates with specific operational changes.
The NPS survey for a paid community should be three questions maximum: the standard NPS question (“how likely are you to recommend this community to a peer in your field?”, 0–10 scale), a follow-up asking for the one thing that most contributed to their rating, and a single follow-up asking what one change would most improve their rating. More than three questions on a survey sent to paying members produces declining response rates with each additional question. A three-question survey with 40% response rate produces more useful data than an eight-question survey with 12% response rate, because the 40% response rate sample is more representative of the full membership.
What to look for in the NPS trend: direction and source. If NPS is declining, the follow-up responses tell you why — common themes in “what most contributed to your rating” are the signal. If NPS is improving, the same follow-up responses tell you which specific changes members attribute the improvement to. The gap between NPS trend and operational metrics trend is the most informative signal. NPS improving while activation rate is flat or declining means the members who are staying are happier, but the members who are leaving are leaving before they reach the survey point that would capture their experience. NPS flat while activation rate is improving means the improvements are reaching new members and keeping them, but not yet producing the enthusiast-level experience that drives high NPS scores and referrals.
For the annual review summary, report the NPS at four quarterly points, the average follow-up themes from detractors (score 0–6), and the average follow-up themes from promoters (score 9–10). The passives (score 7–8) in a paid community are members who are not at risk of churn and are not generating referrals; they are stable and neither the primary problem nor the primary opportunity in a one-year review cycle.
8. Layer 6: economics
With the five preceding layers as context, the economics layer becomes interpretable rather than just descriptive. The three numbers that matter for the annual economics review are: LTV by cohort, CAC trend, and MRR versus target.
LTV by cohort. For each join-month cohort, calculate the total revenue collected from that cohort through the end of the review year, divided by the number of members in the cohort. This is the cohort’s average LTV to date. For cohorts that joined early enough in the year to have 10–12 months of history, this number is a reasonable approximation of their projected LTV — though not a complete one, since some members will continue past year one. The value of the by-cohort calculation is in comparison: cohorts that joined through higher-intent acquisition channels typically have higher LTV to date than cohorts that joined through lower-intent channels, and the difference compounds across the year. If the cohort retention data in layer one showed specific cohorts holding better than others, the LTV data in layer six should confirm that those cohorts produced more revenue per member — and that the acquisition channels that produced those cohorts are worth investing more in.
CAC trend. Customer acquisition cost is total acquisition spend divided by new members acquired in that period. For most paid communities at this scale, acquisition spend is primarily operator time (valued at an hourly rate) plus any paid promotion. Track CAC quarterly rather than annually to see the trend: a rising CAC trend alongside a stable or declining LTV trend is the signal that acquisition is becoming economically unsustainable. A flat CAC trend alongside an improving LTV trend means the community’s unit economics are improving — each new member acquired costs the same but produces more value over their lifetime. The latter is the target: reduce churn, increase LTV, hold CAC flat.
MRR versus target. The final check: the actual MRR trajectory against the target set at the start of the year, and specifically which months showed the largest deviations from target in either direction. Months where MRR outperformed target are worth understanding — was it an acquisition spike, a churn reduction, or a plan upgrade wave? Months where MRR underperformed are worth understanding in the context of the operational data from layers one through five. A month with below-target MRR that also shows below-target activation in the activation rate trend is a month where a specific onboarding or acquisition problem expressed itself economically. A month with below-target MRR that shows no operational anomaly in layers one through five may reflect an external factor — a competing platform launch, a seasonal acquisition slowdown — rather than an internal problem the operator can address.
9. What to decide from the review
The three types of decision a paid community annual review should produce are: what to fix, what to stop, and what to scale. These categories are not interchangeable, and a review that produces only one type of decision is incomplete.
What to fix comes from the diagnostic findings in layers one through four: a specific tenure stage where retention drops, an activation rate trend that is declining, channels with consistently low engagement multipliers, a rising moderation load. Each of these findings should map to a specific operational change with an owner and a timeline. Vague resolutions (“improve onboarding”) do not produce different outcomes in the coming year; specific changes (“rewrite the Day 3 nudge to include the one action most correlated with month-two retention based on the activation cohort data, ship by March”) do.
What to stop comes from the content engagement review in layer three and the moderation load review in layer four. Channels with engagement multipliers consistently below 1 should be reduced or eliminated. Programming types that generate low attendance and low follow-on discussion should be replaced with formats that work better. The annual review is the right time for these cuts, because the full-year data makes the pattern clear in a way that any single month’s data cannot. Stopping something that is not working frees operator bandwidth for the things that are, and in a paid community where the operator’s time is the primary production input, bandwidth is the binding constraint.
What to scale comes from the findings that are working: acquisition channels that produce high-LTV cohorts, onboarding changes that improved activation rate, programming formats with high engagement multipliers, tenure stages where members are staying at high rates. The annual review is the right moment to identify these bright spots and commit resources to amplifying them rather than spreading investment evenly across all activities regardless of their contribution to the outcomes that matter. The Foothold community health check gives operators a baseline score across all six of these dimensions so the annual review can be compared against a consistent benchmark year-over-year, not just against prior-year self-assessment.
The most common failure mode in paid community annual reviews is completeness without consequence: the operator produces all six layers, notes the patterns, and then continues doing exactly what they were doing. The review is only useful if its outputs are decisions that change what the community does in the coming year. A one-page decision document — three bullets under each of the three categories above, with owner and timeline for each — is the right output. The data goes in the appendix. The decisions go in the first page.
Frequently asked questions
What should a paid community annual review include?
A complete paid community annual review should include six layers in this sequence: cohort retention rate by join month (which cohorts are holding versus dropping, and whether the problem is with acquisition, onboarding, or programming at specific tenure windows); activation rate trend across the year (whether the percentage of new members who cross the activation threshold improved, degraded, or held flat); content engagement by channel (which channels are producing discussion versus which are being posted into without response); moderation load per 100 active members (whether incidents per active member are increasing as the community scales); member survey NPS trend (whether members’ reported satisfaction is moving in the same direction as your operational metrics); and economics (LTV by cohort, CAC trend, MRR versus target). The sequence matters: reviewing economics before you understand retention and activation produces conclusions that cannot be acted on.
How do you measure cohort retention rate for a paid community?
Produce a table with join month as rows and months-since-join as columns. Each cell is the percentage of that cohort still active — defined as at least one substantive public contribution in the preceding 30 days — at that tenure point. The patterns to look for: consistent steep drops at months two through three across all cohorts (an onboarding problem); steep drops at months five through eight in specific cohorts only (a programming problem at a specific calendar period); and all cohorts holding flat through month six then declining at months eight through twelve (a long-term value problem at advanced tenure). Each pattern implies a different intervention. The paid community metrics dashboard guide covers how to produce this table from Slack export data and your payment processor.
How do you calculate activation rate trend for a paid community?
For each calendar month, calculate: activated new members divided by total new members. Activation must be defined consistently across all months — a common threshold is two or more public messages plus attendance at one live event, all within the first 30 days. Plot the 12 monthly rates in chronological order and look for the direction. A declining trend means each successive cohort is activating at a lower rate: look for whether acquisition channel mix changed (lower-intent channels produce lower activation rates) or whether the onboarding sequence itself degraded. A stable or improving trend after a specific onboarding change confirms the change is working. The activation rate trend is the leading indicator for the retention curves in layer one.
When should a paid community do its annual review?
January, covering the prior calendar year — but only if the community has at least 12 months of operation and at least 20 new members per month for the cohort table to show distinguishable patterns. Communities that launched mid-year should do their annual review at the 12-month launch anniversary, not at January 1 on partial data. The review should always be timed to a planning cycle where its outputs can be acted on. The most common failure mode is a review that produces findings but no decisions: the three outputs should be what to fix, what to stop, and what to scale in the coming year, each with a named owner and a timeline. Without those, the review is complete but inconsequential.