Paid community retention

He had 3.9% monthly churn and 65% bilateral activation. He rewrote his Day-0 DM twice. Churn didn’t move. A peer showed him where to look. Named concept: the tenure decomposition diagnostic.

Oluwafemi Bankole’s activation numbers looked fine. His overall monthly churn was 3.9%. He had spent three months optimizing his onboarding sequence — rewriting the Day-0 DM, reformatting his intro channel, adding a Day-5 follow-up for stalled members. Activation moved from 62% to 65%. Churn stayed at 3.9%. He had run out of obvious things to try. At a November 2025 operators roundtable, a peer asked him to break the 3.9% figure into tenure bands. The decomposition took twenty minutes. His founding cohort — members in months seven through ten — was churning at 5.8% monthly. His newest members were churning at 2.0%. He had spent three months improving the band that was already the healthiest. Named concept: the tenure decomposition diagnostic — the process of breaking an overall monthly churn rate into tenure-band components to identify which band carries the binding churn rate, so the operator knows which class of intervention to apply rather than iterating on the most recent recommendation.

The community at month ten

Oluwafemi launched The B2B SaaS Partner and Channel Ecosystem Network in January 2025 — a paid Slack community for Channel Sales Leaders, VP Alliances, Directors of Partnerships, and Partner Account Managers at B2B SaaS companies with $15M–$100M ARR that were building or scaling an indirect sales motion. The community was organized around the specific challenges of that transition: identifying which partner types to prioritize at a given ARR stage, designing partner incentive structures that produce active sourcing rather than passive referrals, managing the operational complexity of a partner pipeline alongside a direct sales team, and building a partner success function that does not simply replicate the customer success playbook.

The membership fee was $99 per month. By November 2025 (month ten): 240 active members, $23,760 MRR. 65% bilateral activation — meaning 65% of all members who had ever joined had both posted and received a substantive reply from a peer they had not previously known, within their first seven days. Monthly churn: 3.9%.

Both numbers were below the informal benchmarks he tracked from the operators forum he participated in. Bilateral activation at 65% was 5–10 points below the communities he considered peers. Monthly churn at 3.9% was about 1.5 percentage points above his nearest comparable peer. Neither number was alarming on its own — the community was growing, net member additions were positive, and MRR was trending upward. But the gap between his numbers and peer benchmarks had not closed despite three months of focused work on his onboarding sequence, and he had no clear diagnosis of where the problem was or what to try next.

Three months of work aimed at the wrong population

His assumption going into the optimization sprint was that the churn was coming from activation failures. The logic was intuitive: if bilateral activation was at 65%, 35% of members were either failing to activate at all or activating without receiving a reply. Those members were the most obvious churn risk. If he could improve activation by 5 to 10 percentage points, the downstream effect on churn should be meaningful.

He spent three months on the activation sequence. First rewrite of the Day-0 DM: warmer opening, removed the goal-track question (which he felt read as a survey), replaced it with a single direct ask to post in #intros. Bilateral activation: unchanged. Second rewrite: restored a version of the goal-track question (three embedded options rather than an open text field), added a direct link to the most recent active thread in each goal-track channel. Bilateral activation: 62% to 65%. He also tried a new intro channel format — a weekly pinned prompt thread to give new members a structured entry point — and added a Day-5 follow-up DM for members who had not posted by Day 5. Neither produced a churn improvement.

By November 2025, he had spent roughly 25 to 30 hours on onboarding iteration — copywriting, testing different DM structures, reviewing response rates week over week. The activation rate had moved 3 percentage points. The churn rate had not moved at all. He posted about the experience in the operators forum that he participated in: “I’ve improved activation by 3 points over three months. My churn hasn’t changed. I don’t know if I’m just not doing this right or if activation isn’t actually the driver.”

Chinwe Okoye replied with one sentence: “Can you tell me what your churn rate is in each of your tenure bands?”

The peer

Chinwe ran The B2B SaaS Post-Sales and Customer Outcomes Network — a paid Slack community for VP Customer Success, Heads of CS, and Senior CS Managers at B2B SaaS companies with $20M–$130M ARR managing the operational complexity of a maturing CS function: team scaling, NPS program design, renewal and expansion forecasting, and the transition from reactive support to proactive success management. $109 per month. By November 2025: 310 active members, $33,790 MRR, 71% bilateral activation, 2.1% monthly churn.

Chinwe had run the tenure decomposition diagnostic for the first time in June 2025, at month six of her community. At that point she had 165 members, 2.8% monthly churn, and had been trying to decide between two improvements: (1) better month-two programming (more structured open-problems threads, a monthly cohort virtual coffee) or (2) a peer directory and an annual billing offer. She could not make the decision based on intuition — both options had obvious merit and she had limited time to implement either one properly. The decomposition made the decision unambiguous.

Her June 2025 decomposition:

The shape was back-loaded: the founding cohort was running above the aggregate rate, while the two earlier bands were healthy. The months-1–3 band was at 2.4% — already below the aggregate. The months-4–6 band was at 2.7% — at the aggregate. The founding cohort was at 3.3% — the highest rate and the band with the most members in that stage of the community’s lifecycle.

Chinwe recognized the shape immediately from conversations in the operators forum she had been following. A back-loaded decomposition in a community that did not yet have an annual billing conversion sequence almost always indicated the same thing: founding-cohort members making independent monthly renewal decisions without a specific forward-looking commitment mechanism to interrupt the drift.

She installed the annual billing sequence that month. By November 2025, her founding-cohort band had declined from 3.3% to 1.4% monthly churn, and annual billing conversion had gone from 7% to 71%. Overall churn was 2.1%, down from 2.8%. She had not invested further in her onboarding sequence because the decomposition told her the activation-and-onboarding band was not the driver.

“Run the decomposition,” she told Oluwafemi. “It takes twenty minutes. If your founding cohort is running at something materially above your aggregate, that’s where the work is. If months 1–3 is the elevated band, then yes, activation is the problem and you should keep iterating on the onboarding sequence. But you need to know which one you’re dealing with.”

Running the decomposition

Oluwafemi ran the analysis the following day. The inputs he needed: a list of all current active members with their join dates, and a list of all cancellations in the prior three months with each cancelled member’s join date.

His community used Memberstack for access gating, which stored join dates. His Stripe cancellations had been exported manually each month into a spreadsheet he had been maintaining since launch. He had join dates for 230 of the 240 current members (the other 10 were early-community invitees for whom the join date was approximate); and cancellation dates for all 28 members who had cancelled in August, September, and October 2025.

For each cancelled member, he calculated tenure at cancellation: the month they cancelled minus the month they joined. He grouped active members into four bands: months 1–3 (joined August–October 2025), months 4–6 (joined May–July 2025), months 7–10 (joined January–April 2025, the founding cohorts), and months 11+ (which did not exist yet — the community was 10 months old).

His November 2025 decomposition:

The shape was stark. His months-1–3 band — the band he had been optimizing for three months — was running at 1.8% monthly churn. It was the healthiest band in the community. His founding cohort, the 120 members who had been in the community for seven to ten months, was running at 5.8% — more than three times the rate of his newest members, and 148% of the aggregate rate. Of his 28 total cancellations in the prior three months, 21 had come from that founding cohort.

His activation work had been targeting a band that was already running below the aggregate rate. Everything he had built to improve the first week of a new member’s experience had produced exactly the results it was designed to produce: 65% bilateral activation and 1.8% monthly churn in months 1–3. The community’s onboarding was working. The founding cohort was not.

Reading the shape of the decomposition

The tenure decomposition diagnostic is not just a measurement tool. The shape of the decomposition — which band carries the elevated rate and by how much — predicts which class of intervention addresses the problem.

There are three primary shapes, each with a characteristic cause and a corresponding intervention type:

Front-loaded (months 1–3 binding)

When the months-1–3 band runs materially above the aggregate — typically more than 1.5 percentage points higher — the problem is in the early-lifecycle programming. Members are churning in their first three months at a higher rate than any other tenure group, which means the activation and onboarding sequence is not producing the connection events that predict renewal. The leading indicator is bilateral activation rate: if fewer than 55% of members post and receive a substantive reply within seven days, the months-1–3 band will almost always show elevated churn. The interventions are the Day-0 DM structure (specifically the goal-track question that enables conditional routing), the reply-gap audit (a weekly review of zero-reply intro posts), and the conditional Day-3 nudge that fires only to stalled members and routes to an active thread in the goal-track channel. Front-loaded shapes are the most common early finding for communities in their first six months; they typically resolve as the activation programming matures.

Middle-loaded (months 4–6 binding)

When the months-4–6 band is the elevated outlier, the problem is the gap between the end of the onboarding sequence and the month-three renewal decision. Members who activated successfully in month one found the community useful, and then the programming dropped off. By month four — the first billing date after the initial activation energy has dissipated — they are on month-to-month subscriptions without a specific recent reason to continue. This is the second-month cliff pattern. The intervention is month-two programming: a conditional Day-10 follow-up DM routing activated members to the most active recent thread in their goal-track channel; a monthly open-problems thread seeded by the operator; a monthly cohort virtual coffee for members who joined in the same calendar month. The Day-28 value-crystallization check-in — a DM asking “Since you joined, what’s the most useful thing you’ve found here?” — is particularly effective for the middle-loaded shape: members who cannot name something specific at Day 28 renew at dramatically lower rates than those who can, and improving the Day-28 prompt structure recovers a portion of the months-4–6 churn even when activation looks healthy. The goal-track-keyed Day-75 month-three renewal conversation is typically the highest-leverage single intervention for this shape: data from the month-three renewal framework shows that goal-track-keyed messages with a named value moment and a peer introduction offer produce 30–35 percentage point higher renewal rates than generic check-in messages.

Back-loaded (months 7–12 binding)

When the founding-cohort or months-7–12 band is the elevated outlier, the problem is what happens after the initial activation and month-three renewal programming ends. Members who made it to month seven have done everything right: they activated, they found value in the community, they renewed three times. But they are still on month-to-month subscriptions, and by month seven, eight, nine, and ten, each renewal is an independent decision. The community’s programming may not have evolved with where they are, or it may not have provided them with a fresh, specific reason to stay recently. Without a forward-looking commitment mechanism, the month-to-month decision eventually produces a cancellation that neither the activation sequence nor the standard 21-day at-risk flag was positioned to prevent.

Oluwafemi’s shape was back-loaded. His founding cohort at 5.8% monthly churn was not a programming-quality problem or a member-experience problem. It was a commitment-structure problem: 91% of his founding-cohort members were still on month-to-month subscriptions because he had not installed an annual billing offer or a proactive directory-introduction sequence to precede it.

What a back-loaded decomposition diagnoses

Chinwe asked Oluwafemi one question after he shared the results: “What percentage of your founding-cohort members have converted to annual billing?”

He checked. His annual billing tier ($990 per year, the equivalent of two free months) had been live since October 2025 — one month before the diagnostic. He had sent a brief announcement to all existing members when he launched it. Nine of 120 founding-cohort members had converted: 7.5%.

“That’s the same number,” Chinwe said. “5.8% monthly churn in the founding band and 7.5% annual billing conversion are describing the same situation. Ninety-two percent of your founding-cohort members are making a fresh decision every month about whether to stay. And 5.8% of them per month are deciding not to.”

The mechanism is straightforward. A monthly subscriber at month seven is not thinking about whether the community is worth joining — they already joined. They are thinking about whether it is worth continuing. That is a different question, and it is structurally easier to answer negatively, because the alternative to cancelling is not “miss out on this community” but “pay $99 again this month for something I may or may not have found useful since the last time I consciously thought about it.”

An annual billing commitment changes the question. A member who has committed to twelve months is not evaluating the community monthly; they are evaluating it at renewal time, once per year, from the accumulated weight of twelve months of experience. At that evaluation, the community needs to have been valuable for a significant fraction of those twelve months — not just the month that happened to include the renewal reminder. The commitment structure changes what is being evaluated and reduces the frequency of evaluation to once a year.

But the commitment requires a specific precondition: the member needs a concrete, forward-looking reason to believe the community will be valuable for the next twelve months, not just a backward-looking assessment of the past six. Without a recently facilitated peer connection — a specific person they expect to continue learning from — the forward-looking question is answered from optimism rather than evidence. Optimism, at the moment of an annual billing offer, produces 15–20% conversion. Evidence — a named peer they are actively in conversation with — produces 55–65% conversion, as documented in the directory prerequisite framework.

Chinwe’s installation in June 2025 had followed that sequence: Day-120 proactive operator-facilitated directory introduction (a specific named peer match, sent to every member who had reached month four without an active peer connection) followed 75 days later by the Day-195 annual billing offer structured as a forward-looking question that referenced the peer connection by name. Her annual billing conversion had gone from 7% to 71% in five months. Her founding-cohort band had declined from 3.3% to 1.4% monthly churn over the same period.

Oluwafemi’s implementation

He implemented the two-step sequence starting in December 2025.

The first step was the Day-120 proactive directory introduction. His peer directory had been built in October 2025 — a Notion database with profiles for every member, populated as part of the Day-0 DM checklist (members who completed their profile were routed to two specific peer profiles in their goal-track channel). But he had not been making proactive operator-facilitated introductions: members who browsed the directory did so independently, and he had not been sending individual DMs to match members who had not yet made a bilateral connection beyond their first week.

He started with the founding-cohort members who had no active peer connection recorded in his tracking sheet: 61 of 120 founding-cohort members met that criterion. For each, he reviewed their goal-track selection from their Day-0 DM response and their company stage from their profile, and identified the most relevant other member currently active in the same goal-track cluster. He sent each of the 61 members a personal DM over three weeks in December 2025 and January 2026, introducing a specific named peer and naming the specific challenge each was working on.

Acknowledgment rate from the introduction DMs: 88% (54 of 61). Rate of substantive async exchange within 30 days: 69% (42 of 61). Of the 61 founding-cohort members who received an introduction, 7 cancelled anyway in December and January — a 11.5% cumulative cancellation rate over those two months, compared to the prior three-month rate of 5.8% monthly (17.4% cumulative). The introduction DMs reduced cancellations in that cohort even before the annual billing offer arrived.

The second step was the Day-195 annual billing offer. For founding-cohort members who had received a directory introduction 75 or more days prior and had not yet converted to annual billing, he sent the offer in February 2026. The DM structure he used:

“Hey [Name] — I wanted to follow up on the connection I made between you and [Peer Name] back in December. I saw you had a couple of exchanges about [specific topic from their goal-track] — I hope that’s been useful. I’m reaching out because I want to offer annual billing to founding members: $990 for the full year, which locks you in at the current rate and saves two months. The question I’d ask yourself is simpler than the math: is this community a consistent resource for the work you’re doing over the next twelve months? If you’re working through [goal-track challenge] and finding it useful to have [Peer Name] and others at a similar stage around, I think the answer is yes. If the timing isn’t right, no action needed.”

The offer went to 97 founding-cohort members (the 61 who had received directory introductions in December–January, plus 36 more who had existing peer connections from their first week but had not yet been offered annual billing). Annual billing conversion from that group: 54 of 97 (55.7%). Among the 54 who had received operator-facilitated introductions in December–January specifically, the conversion rate was 62%.

Among the 23 founding-cohort members who did not convert to annual billing (excluding the 7 who had already cancelled), 9 cancelled before April 2026 and 14 remained on month-to-month subscriptions. The 14 who stayed were primarily founding members with active peer connections who had not yet found the annual commitment appropriate for their current situation — several mentioned they were evaluating whether to stay at their company for the full year before making the commitment.

Secondary effect of the introduction DMs: 11 of the 42 founding-cohort members who had substantive exchanges through the December–January introductions posted their first question in a goal-track channel in more than three months. Their engagement had been read-only before the introduction — not quite at the goal-track decay signal stage, but trending that way. The specific new peer connection gave them a concrete reason to bring an active problem to the community that they had not had before.

Results

By April 2026 (five months after the diagnostic, four months after the first directory introductions, two months after the annual billing offer):

The bilateral activation improvement was minimal — 65% to 68% — because he had made no further changes to his onboarding sequence. The improvement came from the same directional work he had done over the prior three months (the Day-5 follow-up and the second Day-0 DM rewrite), which continued to compound slightly. The churn improvement was entirely from the founding-cohort intervention: the directory introductions and the annual billing offer addressed the specific problem the decomposition had identified.

What he had spent three months optimizing — the months-1–3 band — had declined from 1.8% to 1.6% monthly churn. A real improvement, but a small one, because the band was already the healthiest in the community. The founding cohort, which had been generating 75% of his total cancellations in the diagnostic period, was now generating approximately 40% of a much smaller total.

The named concept: the tenure decomposition diagnostic

The tenure decomposition diagnostic is the process of breaking an overall monthly churn rate into tenure-band-specific components, identifying which band carries the binding churn rate, and reading the shape of the decomposition to determine which class of intervention addresses the actual driver.

The diagnostic is valuable because overall monthly churn is a weighted average across all tenure bands, and the interventions appropriate for one band are not effective for another. An operator who has a back-loaded decomposition and invests three months improving their activation sequence will improve the front band slightly — that band was already healthy, and the improvement has a small marginal effect — while leaving the binding band untouched. The churn rate will not move because the binding band is not the target of the work.

The decomposition is not a continuous monitoring tool. Running it once per quarter is sufficient to identify whether the shape is changing — a front-loaded shape that becomes middle-loaded after three months of activation work is evidence that the activation intervention succeeded and the new binding problem is in months 4–6 programming. A shape that stays back-loaded for two consecutive quarters despite a new annual billing offer may indicate a goal-track decay problem rather than a commitment-structure problem, and the diagnostic points the operator toward the goal re-calibration check-in rather than further annual billing optimization.

The diagnostic also resolves the prioritization decision between improvement options without requiring the operator to guess at which lever is most impactful. Chinwe’s June 2025 decision — month-two programming versus annual billing sequence — was resolved by the decomposition in 20 minutes. Oluwafemi’s three months of onboarding iteration could have been redirected after a single decomposition run in October 2025, before the work began.

The specific steps:

1. Export active members with join dates and all cancellations in the prior three months with join dates.
2. Calculate tenure at cancellation for each cancelled member.
3. Group active members and cancellations into four bands: months 1–3, months 4–6, months 7–12, months 13+.
4. Calculate monthly churn per band: (cancellations in that band over 3 months) ÷ (average active members in that band) ÷ 3.
5. Identify the binding band: the band with the highest rate relative to the aggregate. If no band is more than 1 percentage point above the aggregate, the churn is evenly distributed and the intervention priority is programming quality generally, not a specific lifecycle stage.
6. Read the shape: front-loaded → activation and onboarding; middle-loaded → month-two cliff and month-three renewal; back-loaded → annual billing conversion and founding-cohort commitment structure (or goal-track decay if annual billing conversion is already above 35%).

Foothold’s Day-7 scorecard captures each new member’s goal-track selection and activation status. The tenure decomposition can be layered onto the scorecard data at community sizes where member counts make the band calculations stable. If you want to see your community’s tenure decomposition alongside activation and 90-day retention data, the waitlist is below. We’ll show you the shape in the first two cohorts.