Paid community retention

He had been running the community for fourteen months. Bilateral activation was 71%. Monthly churn was 3.1%. A peer asked him to segment the churn by tenure cohort. The number that came back was not 3.1.

Obi Okafor had built the full infrastructure. The activation numbers were solid, the programming was attended, the price raise had landed without a spike. Monthly churn was 3.1% and had not moved in three months. He knew comparable communities were running at 2.1%–2.5%, and he could not find the constraint by looking at the top-line number. Saoirse Donovan ran 2.2% monthly churn on 310 members. She asked one question at a London dinner that located the problem in eight minutes.

The community and the metrics that seemed healthy

Obi launched The B2B SaaS Founders & Operators Network in October 2024 — a paid Slack community for founders and early operators at B2B SaaS companies doing $1M–$15M in ARR. The membership fee was $99 per month. The community was for people navigating the specific inflection points that come in the $1M–$15M range: the first head-of-sales hire, the transition from founder-led selling, the shift from product-market fit to go-to-market, the organizational design questions that arise when a team of twelve becomes a team of forty. The community’s value proposition was peer access — other founders who had recently solved the exact problem the new member was currently facing.

By December 2025, month fourteen: 252 members, $24,948 MRR. He had followed the programming investment ladder deliberately from the beginning. Foothold Day-0 DM with a goal-track question (three tracks: go-to-market and sales; hiring and team; product and metrics), weekly reply-gap audit, conditional Day-3 routing, Day-10 conditional follow-up, Day-28 value-crystallization check-in, Day-75 renewal conversation. He had added the reply multiplier in month eight: three long-tenured members who had already been behaving as stewards without any formal role, invited privately to help close the reply gap in #intros on weeks when his own response time slipped.

Programming: a monthly Founders Roundtable (a 90-minute session with a guest founder or operator who had recently navigated the transition his members were currently in — typically a seed-to-Series-B CEO, CRO, or VP of Sales walking through a real decision with real numbers), a weekly #metrics-check channel (an anonymous benchmarking channel where members could post their current MRR, churn rate, or CAC with a brief question, and receive peer responses without attaching the numbers to their public identity), and a quarterly networking dinner for members in the same city.

In October 2025, he raised the price from $79 to $99 using the three-phase approach from the price raise conversation framework: segmenting the base by tenure and engagement, sending individual DMs to Tier A and Tier B members before any mass announcement, offering annual billing lock to twelve-month-plus members. Churn during the price-raise window: 3.4%, below his 3.2% pre-raise baseline.

By December 2025: bilateral activation 71%, month-three renewal 78%, monthly churn 3.1%. He had also launched a peer directory in October 2025 (month twelve) — a Notion database with fields for role title, company stage, current focus area, and intro preference. Overall directory completion rate: 64%.

Annual billing conversion at month six — the percentage of members who, when offered a discounted annual subscription at the end of their sixth month, chose to pay upfront for a year — was 31%. He had tracked this number since month eight but had not identified it as the constraint. He thought of it as a signal of member satisfaction: members who were confident in the community’s value converted to annual billing; members who were uncertain did not. He had not yet asked whether the 69% who did not convert were churning at a different rate than the 31% who did.

The plateau and what it meant

Three months at 3.1% — October, November, and December 2025 — with no downward trend. He had added the peer directory in October, had run the price raise, had continued the monthly Founders Roundtable with consistently strong attendance (members in their first six months attended at 64%; the overall community attendance rate was solid). The activation infrastructure was performing at 71% bilateral. Month-three renewal at 78% was the highest it had been.

He could identify individual churn events by reviewing cancellations manually. When he did, the reasons were heterogeneous: company sold, founder took a job, “not getting enough value right now,” “found what I needed and will probably rejoin in a year.” No single explanation dominated. He had spoken to four churned members directly in November; two had found the community genuinely useful but felt the ongoing fee was hard to justify when they were in a low-activity phase. The other two had not engaged meaningfully since month two.

The best operators in comparable communities — people running Slack communities for founders and operators in adjacent niches, whom he met at community-building events and spoke to over DMs — were consistently running at 2.1%–2.5% monthly churn. Some were at 1.8%. He was not finding a clear explanation for the gap from his top-line numbers. Everything that was supposed to move churn appeared to be in place and appeared to be working.

He flew to London for a two-day paid-community operators gathering in January 2026. He had come specifically to describe the plateau to operators who had seen it before.

What Saoirse asked at the London dinner

Saoirse Donovan ran The B2B SaaS Growth Operators Network — a paid Slack community for growth practitioners, marketing leaders, and demand-generation operators at B2B SaaS companies in the Series A to Series C range. The fee was $89 per month. By January 2026: 310 members, $27,590 MRR, 2.2% monthly churn, 71% annual billing conversion. She had launched in October 2023 — twenty-six months before the London gathering. She was the operator Obi most wanted to talk to.

He described the situation over dinner on the first evening: the full infrastructure build, the activation numbers, the programming, the price raise. Monthly churn 3.1%, three months at the same rate, no hypothesis that fit the data.

She listened. She did not ask about the programming or the Day-75 renewal conversation or the reply multiplier. Her first question was: “Have you decomposed that 3.1% by tenure cohort?”

He said he had not. He tracked overall monthly churn as a single number.

“Then you don’t actually know where the problem is,” she said. “The 3.1% is a weighted average. It might mean every cohort is churning at 3.1% — in which case the problem is spread evenly across the lifecycle and you need to look at all of it. Or it might mean one cohort is churning at 5% or 6% and the rest are well below 3%, and the one bad cohort is pulling the average up. Those two situations call for completely different responses. If you don’t disaggregate it, you can’t know which situation you’re in.”

She walked him through the decomposition she had built for her own community at month thirteen, and what it had revealed. He took notes. When she finished, he opened his laptop at the dinner table and started pulling the data.

The second-year operator audit

Saoirse described the protocol she had developed after running the decomposition on her own community and then helping three other operators run it on theirs. She called it the second-year operator audit.

The premise of the audit is that by year two, a community with a well-built activation infrastructure produces a characteristic churn pattern that is bimodal rather than uniform. Months one through three carry low churn — the activation infrastructure is converting new members into engaged participants, and members who survive the second-month cliff renew at high rates. Months thirteen and older carry low churn — long-tenure members have formed genuine community relationships, have an established usage pattern, and have accumulated enough history with the community that the cost of leaving is real. The elevated churn is concentrated somewhere in the middle. Where exactly depends on which layer of the programming investment ladder is the binding constraint for that specific community.

The overall monthly churn rate masks this structure. A community running 3.1% monthly churn might be running 1.8% in months one through three, 2.7% in months four through six, 4.2% in months seven through twelve, and 2.0% in months thirteen and older. The arithmetic produces 3.1%. The constraint is in the months-seven-to-twelve band. But an operator looking at the 3.1% overall number has no way to see that without disaggregating.

The audit has three steps:

1. Cohort decomposition — build a tenure-segmented churn table (months 1–3, 4–6, 7–12, 13+) and identify which band carries the binding churn rate.
2. Layer verification — for the binding tenure band specifically, verify which layers of the programming investment ladder are nominally installed versus actually working. A layer is “nominally installed” if it exists in the operator’s system. A layer is “actually working” if members in the binding cohort are engaging with it at rates that correlate with retention.
3. Constraint identification — identify the single sequencing or density issue producing the elevated churn in the binding cohort. In most year-two communities, the constraint is not a missing layer. It is a layer that exists but whose timing, density, or sequencing relative to an adjacent layer is preventing it from producing lift for the specific cohort that needs it most.

“The discipline of the audit,” Saoirse said, “is to run all three steps before you conclude that you need to add something new. Year-two operators almost always respond to elevated churn by adding a new event or a new channel or a new feature. The audit almost always reveals that the problem is in the interaction between two things that already exist.”

Step one: cohort decomposition

Obi exported cancellations from Stripe for the prior six months — July through December 2025. He recorded the join date and cancellation date for each cancelled member. The tenure at cancellation was the number of months between join and cancellation. He grouped them into four bands: months one through three, months four through six, months seven through twelve, months thirteen and older.

For each band, he calculated the average active member count over the same six-month period using Stripe’s subscriber counts at the midpoint of each month. The monthly churn rate for each band: total cancellations in the band divided by average active members in the band divided by six months.

He had the table built by the time the dinner finished.

The 3.1% overall was the weighted average. Months one through three at 1.8%: the activation infrastructure was working. Months thirteen and older at 1.9%: long-tenure members were retained. The binding constraint was the months-seven-to-twelve band at 4.2%.

He showed Saoirse the table. She looked at it for a moment. “This is the most common pattern I see. The activation is good, the long-tenure retention is good, and there’s a specific window in the middle where members are churning at more than double the rate of the cohorts on either side. That window is almost always seven through twelve, and the reason it’s almost always seven through twelve is that it’s the first tenure band that isn’t protected by the discovery phase and isn’t yet protected by the relationship depth that long-tenure members have accumulated.”

She referred to the member directory effect: months one through five, new members are in the discovery phase — the unexplored community environment produces reliable high engagement. By month six, the discovery phase is over. “What you’re seeing in the 4.2% is the discovery phase ending without the infrastructure proposition taking over. The question is which infrastructure is failing them.”

Step two: layer verification for the binding cohort

She walked him through a targeted layer check for the months-seven-to-twelve cohort specifically — not for the community overall, but for that cohort.

The first check: annual billing conversion, disaggregated by what members in that cohort were doing before the month-six offer arrived.

His overall annual billing conversion at month six was 31%. He pulled the cohort data: of the ~72 members currently in months seven through twelve, roughly 32 had converted to annual billing when the offer arrived. The other 40 had not.

Monthly churn for the 32 annual-billing members in the months-seven-to-twelve cohort: 0.9%. Monthly churn for the 40 non-annual-billing members in the same cohort: 6.2%.

“The annual billing conversion is doing almost all of the work in that cohort,” Saoirse said. “Members who converted to annual billing at month six are churning at essentially zero in months seven through twelve. Members who did not convert are churning at 6.2%. Your 4.2% average for that band is a mix of near-zero from the annual-billing members and very high from the non-annual-billing members. The constraint is in why only 31% of members are converting to annual billing at month six.”

The second check: peer directory usage, cross-referenced with the annual billing decision.

He had launched the directory in October 2025 (month twelve of the community). His overall directory completion rate was 64%. But that 64% included members who had joined before October 2025 and received a late invitation, and members who had joined after October 2025 and were invited within their first few weeks.

He segmented by the timing of the annual billing decision relative to the directory launch. Of the members who had gone through the month-six annual billing decision after October 2025 (after the directory existed): 41% had used the directory at least once before the offer arrived. Of those who had used the directory at least once before the offer: 63% converted to annual billing. Of those who had not used the directory before the offer: 16% converted.

Of the members who had gone through the month-six annual billing decision before October 2025 (before the directory existed): 13% converted to annual billing. The baseline for a community without a peer directory.

He read the table twice. The directory was producing a large difference in annual billing conversion for members who used it before the offer arrived. But only 41% of members reaching the month-six annual billing decision were using the directory before the offer. The other 59% were arriving at the decision without a directory experience.

Saoirse said: “The directory is working for members who use it. The question is why only 41% of members use it before the annual billing decision.”

Step three: constraint identification

She asked him to walk through the sequencing: when did a member encounter the directory, in what context, and what was the implicit message about what the directory was for?

He described it: a member who joined after October 2025 received a general invitation to fill out their directory profile as part of the community onboarding process, typically within the first two weeks. The invitation was a brief DM from Obi explaining that the directory existed, linking to the Notion table, and asking the member to add or confirm their profile. Most members who were going to fill it out did so within a week of receiving the invitation.

The directory link was pinned in the community’s channel list. Members could search it any time. There was no prompt specifically for members approaching month six. There was no moment in the community lifecycle where the operator specifically told a member in month four or five: “The directory has members who are working on exactly what you’re working on, and you can request an introduction from me this week.”

Saoirse named the constraint: “The directory exists. Members are invited to fill it out early, and most do. But there’s nothing that specifically prompts them to use it for a concrete purpose in the sixty to ninety days before the annual billing decision. They filled out their profile in week two when the onboarding asked them to, and then the directory passively existed. For a member who hasn’t received a specific introduction request and hasn’t sent one, the directory is infrastructure they know about but haven’t experienced as useful. When the annual billing offer arrives at month six and they ask themselves whether $99 a month is worth paying for another year, the community’s value proposition is still the programming they attend when it is scheduled. The directory is in the list of things the community has, but it hasn’t yet produced anything specific for them.”

She described what she had built in her own community to close this gap: a proactive directory introduction at Day 120 — roughly two months before the month-six annual billing decision for most members, and far enough ahead of the Day-195 annual billing DM that the introduction had time to produce an actual conversation before the forward-commitment decision arrived.

“At Day 120, I look at the member’s goal-track selection from when they joined and their company stage, find one directory entry that matches on those two dimensions, and send them a DM. Not a nudge to use the directory. A specific: ‘I was looking at the directory this week and noticed you and [Member X] are both working on [specific challenge] at a similar stage. Would you like me to connect you?’ The member doesn’t have to search the directory. The operator has already done the search and is presenting the result as a service, not as a homework assignment. The difference in uptake between “go search the directory” and “I found someone specific for you” is very large.”

The second change: move the annual billing offer from Day 165 (month six) to Day 195 (month seven). A thirty-day shift that gives the Day-120 introduction time to produce an experience before the forward-commitment decision. “A member who has had a directory introduction, has exchanged three DMs with the peer, and found the conversation directly useful for something they are working on is making the annual billing decision with a concrete recent memory of what the community infrastructure produced for them. That’s fundamentally different from a member who is trying to abstract the value of the community from a set of sessions they attended and threads they read. The annual billing decision is the member’s answer to: what does this community give me that I couldn’t replicate elsewhere? If the answer is a peer I found through a directory and had a specific useful conversation with, the forward commitment is easy. If the answer is ‘sessions I attend sometimes,’ the forward commitment is much harder.”

What Obi changed and how

He began implementing both changes in January 2026.

For the Day-120 proactive introduction: he pulled the list of all members currently between Day 90 and Day 135 — the cohort approaching the thirty-to-seventy-five day window before the new annual billing offer timing. There were 22 such members. For each one, he reviewed their goal-track selection from their Day-0 DM (from Foothold’s member data), identified their company stage from their directory profile or from posts in the #metrics-check channel, and searched the directory for one entry that matched on both dimensions.

He found a match for 19 of the 22 members — three had goal-track selections or company stages that were underrepresented in the directory at that specific combination. For those three, he sent a different kind of message: a question about what they were currently working on, with an explicit offer to search the directory on their behalf once he knew the challenge. For the 19 with direct matches, he sent the introduction DM: “I was going through the directory this week and noticed that you and [Member X] are both focused on [specific challenge, e.g., ‘the transition from founder-led sales to a first sales hire’] at a similar stage. Both of you are in the $2M–$5M range with a two-person go-to-market setup. Would you like me to connect you? Takes me about two minutes.”

Fifteen of the 19 replied positively within four days. Fourteen successful introductions were facilitated in January — in each case, Obi sent a brief message to the target member first before making the connection, following the approach from the member directory effect: naming the specific challenge, asking if the target was open to a brief conversation, then connecting the two members only after the target confirmed.

He also updated the annual billing offer workflow in Foothold: the trigger moved from Day 165 to Day 195. In the offer DM itself, he added one line that referenced the directory specifically: “You now have access to [N] member profiles in the peer directory — you can request an introduction from me any time for as long as you’re a member. That’s a resource that stays valuable as your challenges change.”

For the members going through the Day-120 introduction in January, the annual billing decision would not arrive until March (Day 195 from their join date). He would not have clear conversion data until April.

In February, he ran the Day-120 proactive introduction for the next cohort of members approaching the window: 18 members, 14 matches found, 13 introductions facilitated. In March, the same: 21 members, 16 matches, 15 introductions.

The results: months-seven-to-twelve retention and annual billing conversion

The data tracked from April through July 2026 — the four months during which the first cohorts to receive both the Day-120 introduction and the Day-195 annual billing offer were making their forward-commitment decisions:

Of the members who converted to annual billing between April and July 2026: 78% had had at least one proactive directory introduction before the offer arrived. Of the members who did not convert to annual billing during the same period: 71% had not had a proactive introduction.

He also tracked the outcome of the January introductions specifically. Of the 14 introductions facilitated in January, he followed up at 30 days and 90 days. At 30 days: 12 of the 14 pairs had exchanged at least one DM. At 90 days: 9 reported that the conversation had been directly useful for something they were working on. Four said the conversation had been useful but not immediately relevant to their current challenges. One did not respond to the follow-up.

He noted the same caveat Adaeze Nwosu had noted when she tracked directory outcomes in her own community: the members who had proactive introductions before the annual billing decision were also probably more engaged with the community generally, and more engaged members convert to annual billing at higher rates regardless of the directory. He could not rule out that selection effects were driving part of the correlation. But the directionality of the data was consistent: the introduction was arriving early enough in the member lifecycle to produce an actual directory experience before the forward-commitment decision, and the forward-commitment decision was changing.

What year two is actually about

The programming investment ladder describes the first-year build: the ordered sequence of retention interventions that a new community operator must install, in sequence, before the community can hold members past the second-month cliff and past the first conscious renewal decision at month three. Year one is about installing the infrastructure in the right order. The most common year-one mistake is skipping to programming — events, channels, speaker series — before the bilateral activation foundation is in place.

Year two is a different kind of problem. The operator who has followed the ladder correctly arrives at month thirteen with a community that activates new members well, retains them past month three, and holds long-tenure members at low churn. The overall churn number looks acceptable — 3.1%, 3.4%, maybe 3.8%. It is not yet at the level of the best operators in comparable niches, but it is not alarming. The natural year-two instinct is to keep adding: a new event format, a new channel type, a new programming element that might produce the incremental lift. This is almost always the wrong response.

Year two is about finding the binding constraint in what already exists, not adding to the stack. The second-year audit works because the overall monthly churn rate, by year two, is almost never uniformly distributed across tenure cohorts. The activation infrastructure has made month one through three genuinely good. The relationship depth of long-tenure members has made month thirteen and older genuinely good. Somewhere in the middle — almost always in the months-seven-to-twelve window, after the discovery phase has ended and before the member has accumulated the relationship depth that makes long-tenure retention durable — there is an elevated rate that is pulling the overall number up.

The operator who looks at 3.1% overall and adds a new monthly event is adding programming to a heterogeneous population. The new event will be attended mostly by members who are already engaged — members in months one through three who are still in the discovery phase, and long-tenure members who have an established community relationship. The members in months seven through twelve who are churning at 4.2% are, in Saoirse’s formulation, past the discovery phase without yet having the infrastructure proposition. A new event does not answer their question. Their question is not “is there something new to explore in this community?” The answer to that question was no by month six. Their question is “is this community infrastructure I need for my professional life in the next twelve months?” A new event is another session they will attend when scheduled. The peer directory, sequenced so that they have a concrete directory experience before the annual billing decision, is infrastructure they use when they need it.

The second-year audit’s discipline is to disaggregate the churn number before asking “what should I add?” and to exhaust the sequencing and density hypotheses before concluding that a new layer is required. In Obi’s case, both changes — the Day-120 proactive introduction and the Day-195 annual billing offer timing — required no new infrastructure. They required better sequencing of the infrastructure that already existed, surfaced by looking at one tenure band’s churn rate rather than the community’s overall churn rate.

If you want the bilateral activation tracking and conditional routing infrastructure that makes year one work — the foundation that the year-two audit checks — Foothold handles layers one through four of the programming investment ladder in a single Slack app install. The peer directory, the Day-120 proactive introduction, the reply multiplier, and the annual billing offer sequencing remain the operator’s work. But the insight that year two is about identifying the binding cohort and the sequencing constraint in the layer that serves it, rather than adding new programming to an already-well-built stack, is the one that most year-two operators miss — because it requires disaggregating a number that looks fine into the tenure-band structure that reveals where it isn’t.