Most member counts are an archive of intentions. They record everyone who once decided to join, and almost nothing about who is here.
That is not a vanity problem. It is that the number merges several completely different groups of people, and the right response to each one is different — including, for most of them, doing nothing.
Nobody un-joins
People leave communities by not arriving, which produces no event and no record, so the list only ever grows.
A year in, a healthy community and a dead one can have identical member counts. The difference is invisible in the number and obvious in the room, which is why owners who watch the count are the last to know. Growth in that figure is not evidence of anything except that your signup page works.
This post is about what to do with the accumulated list. If what you want is to understand why people stopped coming, that is a different job with a different method — this one starts after you have accepted that most of the list is not coming back.
Four groups, not one
The word "inactive" covers at least four situations that have nothing to do with each other.
| Group | What it actually means | What to do |
|---|---|---|
| Never activated | The promise and the first visit did not match | Fix the arrival, not these people |
| Dormant | Something changed, often something you changed | One personal message, once |
| Seasonal | Their year has a shape and it is not yours | Nothing. They come back on their own |
| Dormant and paying | You are being paid for something nobody is using | Decide on purpose, soon |
Only the second and fourth rows deserve any action at all, and the first row — usually the largest by a distance — is a problem with the first session rather than a population to be won back.
The re-engagement campaign backfires
The standard move is an email to everyone who has gone quiet, and it reliably underperforms in a way that is worth understanding.
Those people are not undecided. They made a decision, quietly, some months ago, and the campaign asks them to make it again — except this time consciously, with a button in front of them. You are not re-engaging a dormant member; you are interrupting a settled one and requesting a fresh verdict. Most of them will reach the same conclusion, and now it is on the record.
The second problem is aim. A campaign cannot tell the never-activated from the dormant from the seasonal, so it says the same thing to a person who never saw the place and a person who loved it until something specific went wrong. Neither message lands, because neither was written to anyone.
What does work is one message to one person
A single note to somebody you can actually name outperforms any campaign, and it does not scale, and that is fine.
Pick five people who used to be active. Write to each about the thing they were last working on, because you can see it and they will know you looked. Ask one question. Do not mention the community, the plan, or coming back.
Roughly speaking, this works because it is not marketing — the person receiving it can tell the difference immediately, and so could you. What comes back is not usually a returning member; it is an explanation, which is worth more, and occasionally it is both.
Some of them are just seasonal
A large share of what gets filed as dormancy is a calendar you have not noticed.
Teachers vanish in exam season, accountants at year end, parents over the school holidays, and anyone in a northern country during whatever their summer is. If your members cluster in one profession, your community has a shape to its year whether or not you have mapped it, and the dip that panics you in August is the dip that happens every August.
Before treating quiet as decline, check whether it happened at this time last year — that is nearly the whole diagnosis, and the rest of it is in why communities go quiet.
Should you remove anyone?
For a free community the answer is almost always no, and the reasons people give for yes are usually about the chart.
A dormant free member costs you nothing and carries a small option value: people do come back, months later, because something in their life changed rather than because you sent anything. Deleting them converts a possible return into a certainty of nothing, in exchange for a better-looking percentage.
If the number is misleading you, fix the reporting rather than the list. Removing people so that a ratio improves is a way of lying to yourself with extra steps, and the ratio was the thing you were supposed to be learning from — see the metrics worth tracking for the version that does not require deleting anybody.
The dormant paying member
Somebody who pays every month and has not opened the place since spring is the one case where doing nothing is itself a decision.
Most businesses call that revenue and move on. It is worth being precise about what it actually is: money you will lose all at once rather than gradually, from someone who will feel, correctly, that they were not reminded. It is also where refund requests and card disputes come from, and those arrive with a story attached that other people hear.
An occasional "you have not used this in a while — do you want to pause?" costs real money and buys the only goodwill that survives a credit-card statement. It is not charity and it is not clever retention; it is the difference between a business people trust and one they merely subscribe to, which eventually shows up in what you can charge.
Count returning members instead
If you replace one number today, replace joins with returns.
How many people who were here last month are here this month? That single figure describes the room rather than the funnel, it cannot be inflated by a good week of signups, and it goes down immediately when something breaks — which is exactly what you want from an instrument.
Everything else follows from it. A community with three hundred returning members is a real place regardless of whether the total says four hundred or forty thousand, and the gap between those two numbers is only interesting as a churn question, not as an inventory to manage.
What your admin can actually tell you
You cannot act on any of this without seeing who is in which group, and that is a filtering question.
In Mateflow the member list filters by when someone was last active, in four bands — the past week, the past month, the past ninety days, and inactive — and each member carries a status of active, inactive, suspended or banned. You can save a filter set as a named segment so the same list is one click away next month, and last-active is one of the fields you can include when exporting members, which is how you get a real before-and-after rather than an impression.
One limit is worth knowing: after ninety days everything collapses into a single "inactive" bucket. Somebody four months gone and somebody three years gone look identical there, and they are not the same person — one may still open a message from you and the other almost certainly will not, so treat that bucket as a starting point and use the export when the distinction matters.
Three things not to do
Do not email everyone who has gone quiet at once. It is the least aimed message you will ever send, and it asks several thousand people to consciously confirm a decision they had stopped thinking about.
Do not delete members to improve a percentage. The percentage was the measurement; changing the population changes the reading without changing anything real.
Do not count a dormant paying member as a satisfied one. They are the least satisfied person on your list, and the only reason you have not heard about it yet is that they have not looked at the charge.
The bottom line
An inactive list is not a problem to be solved. It is mostly the residue of a working signup page, and most of it needs nothing from you at all.
Separate the four groups before you act, fix arrival rather than chasing the people it failed, leave the seasonal ones alone, write personally to the handful you can name, and be honest about anyone who is paying and absent. Then stop watching the total, because the number that tells you whether this place is alive is how many people came back — and no amount of pruning, emailing or reporting will move that one without moving the thing underneath it.