Foothold

Content strategy & engagement

My paid community had a 64% activation rate and a weekly engagement rate stuck at 22% for four months. When I categorised my last 20 posts, 13 of them were designed to be consumed — not replied to.

23 October 2026 · 12 min read

Camille Bernard ran The Product Marketing Alliance for nine months before a peer asked her a question she had never thought to answer: what is your average replies per post? The number she calculated — 0.8 — explained four months of a flat engagement rate. This is what she found, what she changed, and what happened in six weeks.

Eight posts a month, a 64% activation rate, a 22% engagement rate

The Product Marketing Alliance is a paid Slack community for senior product marketers and PMM directors at B2B SaaS companies. $139 per month. Camille launched it in March 2025, six months after leaving her VP of Product Marketing role at a Series B SaaS company. The idea came from a problem she had felt personally: at her level, there was nowhere to go to ask real questions about positioning decisions, PMM team structure, or the politics of cross-functional alignment. She built the community she had wanted.

By November 2025 — month nine — she had 240 members and $33,360 MRR. She had rebuilt her onboarding sequence in June 2025 after discovering her week-one activation rate was 31%. The sequence she built in June — a four-question intake form at checkout, a goal-keyed Day-0 DM within an hour of joining, a conditional Day-3 nudge for members who had not yet posted, a Day-7 operator scorecard naming three at-risk members — had moved activation to 64% within two cohort cycles. Monthly churn had dropped from 6.8% to 4.2%.

She had also built a content cadence in month two and maintained it without interruption. Every Tuesday and Thursday she posted to the community. Topic variety. On-theme. Never promotional. Eight posts a month for seven straight months.

Her weekly active poster rate: 22%. Four months ago it had been 23%. Before that, 24%, 23%, 22%. No trend in either direction.

She had expected that fixing activation would lift the engagement rate over time as more activated members accrued in the membership. The activation fix had lifted month-two renewal from 58% to 77%. But the weekly active poster rate — the share of all current members who posted at least once in the past seven days — was not moving. She was doing what the playbooks said to do, and the number that mattered for retention was flat.

The question she had not thought to ask

She brought it to a peer operator she had met through a community-of-communities Slack group in August. They had a 90-minute strategy call. He ran a paid Slack community for revenue operations leaders.

His first question: “What is your average replies per post?”

She did not know. She had been tracking her posting frequency — eight per month, consistent — and the community-level engagement rate, but she had never computed the reply average as a standalone metric. She went to Slack and calculated it on the call: she scrolled through her last 20 operator posts and counted the reply thread counts.

The average was 0.8 replies per post.

He told her: for a 240-member community with 64% week-one activation, you would expect five to eight replies on a well-constructed post that reaches a topic members care about. A 0.8 average means almost none of your posts are producing conversations. They are being seen — probably appreciated — but not replied to. Pull your last 20 posts and categorise each one. There are three types.

The 20-post content ratio audit

After the call, Camille went back through the workspace and classified every one of her last 20 operator-initiated posts into one of three categories:

Consumable. Content that delivers value by being read. Resource shares (link plus two to three sentences of context), announcements (upcoming events, new member welcomes, admin notes), event recaps, or content summaries. The member can extract the full value of the post by reading it and reacting. No reply required.

Directive. Open questions or share-your-win prompts. “What is the biggest positioning challenge you are facing right now?” or “What tool are you using for competitive intelligence?” Content that asks members to volunteer something in response. Requires a reply to produce value — but places the cognitive burden of forming that reply entirely on the member.

Participatory. Positioned takes, stated opinions with invitations to respond, or curated threads with operator framing. Content that gives the member something to react to rather than something to originate. The value is co-produced by the reply exchange, and the format makes replies structurally easy: a member needs only to agree, disagree, or add a qualification to an existing position.

Her 20-post breakdown:

Thirteen of twenty posts — 65% — were consumable. They required no reply to provide value.

Why resource shares do not move the engagement rate

The reaction counts on her resource shares were real. Twelve reactions per post meant members were seeing the content, clicking the links, and signalling value. At least twelve members per post were sufficiently engaged to register a response.

The problem is what the weekly active poster rate measures. It counts the members who posted at least one message in the seven-day window. Not the members who reacted. Not the members who clicked a link. Not the members who read an article. Only the members who typed something and sent it.

Reactions are not posts. A member who hits the :fire: emoji on a resource share has not been counted as an active poster that week. They could have reacted to every post Camille made for the past four weeks and still be in the non-active bucket from the engagement rate’s perspective.

This creates a trap that is easy to miss. Camille had been tracking her engagement rate and seeing it flat. She had also been noticing that her posts were getting reactions. The reactions felt like evidence that the content was working. The engagement rate said something different. She had been misreading reactions as a proxy for posting activity, when they are a proxy for content consumption — a different thing entirely.

The practical consequence: nine resource shares per month, each producing twelve reactions and 0.6 replies, were generating 108 reaction events and contributing approximately 5.4 reply events per month — while not adding a single name to the active poster count unless one of the reactor members also happened to post something else that week. The resource shares were invisible to the metric she was trying to move.

The open question ceiling

The open questions troubled her more than the resource shares. She had added them specifically to drive participation. They were the format that felt most like an invitation to engage. And in months two through four, when the community was smaller and most members knew each other by name, they had worked: the “what are you working on this week?” prompt reliably produced eight to ten replies in a 70-member community.

At 240 members, the same prompts were producing 4.2 replies from the same six people.

The mechanism: open questions work at small community sizes because of social density. In a 60-member community, most members know the other members by name. The operator knows everyone. When a member sees an open question from the operator they have DM’d directly, in a community where they know half the people who will read their reply, not responding feels like ignoring someone you know. The social pressure to participate is real and operates on members who would not otherwise post.

At 240 members, most members are strangers to each other. A member in month three of their membership may know three to five other members by name. When they see “what is the biggest positioning challenge you are facing right now?” they face a different calculation. They have to:

The threshold to overcome these barriers is different for every member. The members who cleared it regularly were the six or seven regulars: members with high social confidence in the space, established relationships with other regulars, and a posting habit that had become habitual rather than effortful. For the other 233 members, the open question was not a low-friction invitation. It was a higher-friction request than it appeared.

Open questions have a community-size ceiling. They work when social density is high enough to supply the social motivation that makes posting feel normal rather than exposed. Past that threshold — somewhere between 100 and 200 members for most communities — they stop pulling in quiet members and serve mainly the posting cohort that was already posting.

Why positioned takes pull in quiet members

Camille’s three positioned takes had averaged 8.7 replies and had produced something the open questions had not: replies from members who had not posted in three to six weeks. Members she had been watching in the non-active bucket.

The mechanism is structural. A positioned take removes all three barriers that stop quiet members from responding to open questions.

No blank-page problem. The member does not need to form an opinion from scratch. The operator has already formed one and stated it clearly. The member needs only to decide whether they agree, disagree, or want to add a qualification. Agreeing is the lowest-friction response in any discourse format: “This tracks with my experience. At my last company we tried to combine CI and win/loss under the same PM and it produced exactly the problem you describe.” Disagreeing is the second-lowest: “I’d push back on one part — I think the real issue is that neither CI nor win/loss has an owner at most companies, not that they’re combined.” Either response is easy. Neither requires the member to construct an opinion from nothing.

No blank-slate risk. A reply to a debate post is structurally invited. The operator has explicitly asked for pushback or additions. A member who disagrees with a positioned take is doing exactly what the post asked them to do. The risk of the reply being unwelcome — the social cost that stops many members from posting into an open feed — is structurally lower for a response to a positioned take than for an unprompted post or a reply to an open question.

No social threshold problem. A positioned take creates a shared object: the position being debated. Members can reply to the post without knowing each other because they are all responding to the same thing. The thread becomes a conversation with a shared subject, not a room full of strangers being asked to introduce themselves. Members who would never post in response to “what is your biggest challenge?” will post in response to “my position is X and here is why — change my mind” because the latter gives them something specific to address that is the same for every member reading it.

The quiet members in Camille’s community had opinions. They were not absent because they were disengaged. They were absent because the format of her posts had not given them a low-friction re-entry point into the conversation. Positioned takes created that entry point.

Six weeks of flipping Tuesdays

Camille’s experiment: for six weeks, she replaced her Tuesday resource share with a positioned take. No other change. Thursday stayed as a curated thread or announcement. Same eight posts per month. Different format on Tuesdays.

She wrote the Tuesday positioned takes in the same 10–15 minutes she had been spending on resource shares. The difference was that instead of finding a good article and writing two sentences of context, she was forming a clear opinion on a topic relevant to the community’s domain and naming it.

Week one: “I don’t think most PMMs should run competitive intel the way analysts do. CI is real-time and decision-enabling; market analysis is periodic and frame-building. Treating them as the same job means you get bad versions of both. The correct move is to give CI to the PMMs closest to the deals, not to a centralised research function. Change my mind.” Eleven replies. Three from members who had not posted in six or more weeks.

Week two: “Annual pricing should not be offered to new members before they have experienced the community. The member who signs up on day one and immediately pays annually is the member who hasn’t yet decided if the community is worth it. You are collecting a year of revenue from someone who has not cleared their first-week activation threshold, which means your renewal-period drop in annual churn is actually a delayed version of the same problem.” Nine replies, including a thread that ran for two days.

Week three: “Ideal PMM-to-PM ratio is 1:3, and I have been wrong about this for six years. I used to argue for 1:5 on the grounds that PMMs should be generalists who can span multiple product lines. What I found at my last company was that 1:3 is the point where PMMs can do deep discovery work rather than coverage work, and the output quality difference is not marginal — it’s the difference between PMMs who know what customers are trying to do and PMMs who know what product is shipping.” Fourteen replies. The thread ran 36 hours.

Week four: “PMMs should attend sales calls, but only pre-deal discovery calls with ICPs — not late-stage negotiation calls or CSM handoffs. Fight me.” Thirteen replies, including a detailed disagreement from a PMM who argued the late-stage calls were where competitive positioning actually mattered most.

Week five: “Launch announcement cadence: one warm-up post, then quiet until launch day. Everything else is rocket-exhaust marketing that trains your audience to ignore launch signals.” Seven replies — the lowest of the six weeks, which she attributed to the topic being less debatable rather than to format.

Week six: “Product-market fit is not a threshold event. It’s a rate-of-change signal in customer retention. If you are asking ‘have we reached PMF?’ you are asking the wrong question — the question is whether your retention rate is improving or deteriorating. PMF is a direction, not a destination.” Twelve replies.

After six weeks:

She also noticed something she had not planned for: the positioned takes were producing threads that became the following week’s easiest curated thread candidates. The week-three discussion on PMM-to-PM ratios produced an exchange between three members that was the clearest explanation she had ever seen of the coverage-versus-depth distinction. She highlighted it the following Thursday with two sentences of framing. Six more replies from members who had missed the original thread.

Participatory content was generating the archive that made curation easy, which was generating additional participatory content from members who had missed the first round. The loop the content strategy reference guide describes was working — but it required the positioned take to start it.

What to do with resource shares

Camille had not wanted to eliminate resource shares entirely. The reactions on them were real. Members were reading the articles she was surfacing and appreciating the curation. She did not want to stop providing that value.

Her solution was a channel change, not an elimination. She moved resource shares from the main Slack feed to a monthly reading list section in her email newsletter to the community mailing list. The resources still reached the same members. They no longer lived in the Slack workspace, where they had been generating reactions without generating reply threads and without contributing to the weekly active poster rate.

The Slack feed became more participatory. The email newsletter became more useful: members who had been getting the resources as Slack posts — easy to miss, buried in the day’s Slack traffic — were now getting them curated and framed in a monthly digest they could read at a time that suited them.

The principle she extracted: content that produces value through consumption belongs in a channel designed for consumption. Email is designed for reading. A dedicated #resources channel with a norm that reactions are the expected response is designed for browsing. The main Slack feed in a paid community is designed for conversation. Routing each type of content to the channel whose affordance matches it means no format is asked to do a job it was not designed for.

The secondary benefit: with resource shares removed from the main feed, the positioned takes and curated threads were more visible. The feed was no longer a mix of things that required different responses; it was consistently a conversation-first space. Members who opened the workspace after a few days away saw positioned takes and thread highlights rather than links and announcements, and were more likely to have something to say.

The 20-post content ratio audit

The tool Camille now runs every six weeks:

  1. Pull your last 20 operator-initiated posts from the workspace. Scroll back through your own message history in your main channel, or use Slack search filtered to your username to find them quickly.
  2. Categorise each post as consumable (resource share, announcement, recap), directive (open question, share-your-win prompt), or participatory (positioned take, stated opinion with invitation, curated thread with operator framing).
  3. Count the average replies per post in each category. The consumable average and the participatory average are the most revealing comparison.
  4. Calculate the percentage of posts in each category.
  5. Compare to benchmarks. Communities with weekly active poster rates above 35% typically have 50% or more participatory content. Communities below 25% typically have 25% or less participatory content and a heavy consumable mix. If your positioned takes are averaging more than four times the replies of your resource shares, the mix is your primary lever.

The audit takes 20 to 30 minutes. Operators who run it for the first time typically find 60–70% consumable content, which is a predictable result: resource shares are easy to write and feel valuable because the articles themselves are good. The fact that they do not produce replies is invisible until you measure it.

For a reference on what each content format is trying to produce and how to build a four-week calendar that maintains the right mix, see the Slack community content strategy reference guide — it covers the three formats and the 15-minute weekly topic harvest that keeps the positioned-take pipeline full. For the broader engagement rate picture and the three-stage diagnostic that tells you whether content is the lever or whether activation is the upstream problem, see the engagement rate guide.

Three months later

Camille’s results at the three-month mark from when she started the experiment:

The improvement in month-three renewal was the most significant downstream consequence. Members who have participated in at least one positioned-take thread within their first three months of membership have a specific experience they can point to when they evaluate whether to renew: a conversation where they contributed something and got genuine responses. Members who have only consumed resource shares and reacted with emojis have a different renewal-time impression — they received value, but they did not participate in creating it. The renewal calculus is different.

The activation sequence she had built in June was necessary. Members who never cleared week-one activation churned before they ever encountered a positioned take. Fixing activation brought members to the threshold where content mattered. The content format ratio determined what happened next — whether they became participants or remained consumers. A community of consumers is not stable at any price point. A community of participants renews.

Related: Slack community content strategy — the three formats and the 4-week calendar · The denominator trap: why the engagement rate falls as the community grows · The activation shadow: members who activated and faded by month three · What the best paid Slack communities do in weeks three and four.