The Ninety-Day Mirror: Reading What Your Member Base Is Telling You

Start With Cohorts, Not Averages

An average tells you how the whole membership behaves at one moment, which mixes together people who joined last week with people who joined two years ago. A cohort groups members by the month or week they joined, so you can follow each group forward through time. This is the single most useful change most community builders can make to their thinking about retention. When you compare the March cohort to the June cohort, you may find that the June group stays active for a shorter period, which points to something that changed in your onboarding, your content calendar, or the kind of people you attracted through a particular channel.

To set this up, list every member with their join date and make a simple table with columns for each month after joining. In each cell, mark whether that member did anything meaningful during that month. Define meaningful in advance and keep the definition narrow. Posting a reply counts, reacting to a post may not, and simply logging in may not count at all. Over several months, the table will show you the shape of retention for each group rather than a single flattened number.

Be careful not to read too much into small groups. A cohort of four people can swing wildly from one month to the next for reasons that have nothing to do with your community. Where possible, wait until a cohort contains enough members that a single departure does not change the picture dramatically.

Track the First Action, Not Just the First Visit

Many communities measure how many new members arrive, but arrival is a weak signal. What matters more is what a member does in their first days. A useful question to ask of each cohort is how many members took a first meaningful action within a set window, such as their first week. That action might be introducing themselves, asking a question, responding to a prompt, or joining a small conversation. The exact action matters less than the fact that you chose one and watched for it consistently.

Comparing first-action rates across cohorts often reveals more than retention figures do. If the first-action rate drops sharply for a particular intake source, the problem may lie with how those people were invited rather than with the community itself. A member who arrived through a broad, loosely targeted link may never have understood what the space was for. A member who was personally invited by a current participant may behave very differently from one who simply found a listing. Separating these paths lets you see which doorways produce people who stay.

Once you know which first actions predict later engagement, you can design around them. If members who introduce themselves in week one are twice as likely to still be active at ninety days, then the introduction is worth protecting, with a clear prompt and a visible place to respond. You are not guessing that introductions matter; you are testing whether they do.

Look for the Quiet Middle

Retention problems often concentrate in a stretch that is easy to overlook: the period after the novelty of joining has worn off but before a member has built any real ties. Members in this stage have seen the basics, they may have posted once or twice, and they have not yet developed a routine or a relationship that would pull them back. Their absence does not look like a crisis because they are not complaining. They simply stop appearing.

To find the quiet middle, look at members who were active in their first week, then went inactive for a stretch, and then either returned or did not. The timing of their drop-off is informative. If many members fade between roughly the fourth and eighth week, that pattern suggests the community offers plenty of entry points but few reasons to keep returning on a regular schedule. The fix is rarely more content. It is usually a repeating structure, such as a weekly thread, a standing call for updates, or a small group that meets on a predictable rhythm, giving members a reason to check back that does not depend on luck.

Ask Members Directly, and Ask the Right Question

Data shows you what happened; it rarely explains why. The fastest way to close that gap is to ask people who have left or gone quiet. The trick is to make the request easy and specific. A broad message asking for feedback will get a handful of responses, mostly from people who are already happy. Instead, write a short note to members who have been inactive for a defined period. Thank them for being part of the space, mention one concrete thing they did, and ask a single open question, such as what would have made it more useful for them.

Do not argue with the answers or try to talk people out of their experience. Record what you hear, group similar responses, and look for patterns across several weeks. Some answers will be about timing, some about the fit between the community and what the person came for, and some about things you can actually change. Keep the list of changes you make small enough that you can see whether they affect the cohort numbers in later months.

A Practical Retention Review Checklist

Use this checklist once a quarter to keep the method grounded. First, confirm that you have a written definition of a meaningful action and that it has not drifted over time. Second, build or update your cohort table so that each month of joiners can be followed forward. Third, compare first-action rates across the intake paths you use, and note which paths produce members who stay past the first month. Fourth, identify the stretch where most departures cluster, and look at what the community offered in that window. Fifth, send a short, specific question to a sample of inactive members and record their responses without judgment. Sixth, choose no more than two changes to test in the next cycle, and decide now what result would count as a success.

Finally, be honest about the limits of this review. A cohort table cannot tell you whether a member is happy, only whether they acted. A small community may never produce statistically clean numbers, and that is acceptable if you rely on the qualitative answers more heavily. The goal is not a perfect measurement. It is a habit of looking at your members as specific people who arrived at specific times, through specific paths, and who made specific choices about whether to stay. That habit, repeated over a few quarters, tends to reveal far more than any single retention figure ever will.