Your own users are a community, and you are ignoring it

You keep a list of the communities you work. A couple of subreddits, a Discord, a forum you post in on Saturdays, a newsletter that took you two months to get into. Each one has a state, a cost in mornings, and a running argument about whether it earns its place.

The community with the highest return per hour you have ever worked is not on that list. It is the people who already pay you.

The one community that already chose you

Every community on your list is full of strangers who have not decided anything yet. Your userbase is the one group that has already made the decision the communities exist to produce. They know what the product is, they have been through onboarding, and a meaningful fraction of them are paying.

They are also the only community you can lose overnight, and the only one where the loss is invisible until it has already happened. A subreddit that stops working announces itself: the clicks drop. A userbase that stops working looks exactly like a userbase that is working, right up until the month the churn number moves.

Almost nobody gives it the same attention as a subreddit they posted in twice. There is a reason for that, and it is not laziness. Retention work has no room to walk into. There is no feed, no moderator, no obvious next action. So it loses every session to whichever community feels loudest that morning.

Seven things to do, on a schedule

What retention work needs is the same thing community work needs: a trust list, so it does not depend on remembering. Seven of them, with the cadence each one actually deserves:

Loop	Cadence	What it surfaces
Activation gap	weekly	Signups who never reached first value
Cancellation reasons	on cancel	One required question
Five conversations	monthly	Churn causes that never appear in data
Cohort retention	monthly	By signup month, never a rolling rate
Ship-and-tell	on release	The cheapest anti-churn action there is
Dormancy	weekly	Paying but not using
Expansion	quarterly	What a satisfied customer buys next

Four of those are worth stating properly, because they are the ones people get wrong.

People who signed up and stalled are your easiest sales

These are people who signed up and never reached the point where the product does the thing. They already chose you. They got past the landing page, past the objection, past the form. They then hit something that stopped them, and you have no idea what it was.

Nothing in acquisition is cheaper than this. A new signup costs you a morning in a community. An activation-gap signup costs you an email.

Group people by the month they joined, never a rolling churn rate

A single churn percentage averages together people who joined last week and people who joined last year, and the average hides the shape. Group by the month someone signed up and follow each group forward, and the shape becomes obvious.

The specific thing this catches is churn-by-design: a product that solves a one-time need, sold on an annual plan. Every cohort retains beautifully for eleven months and then leaves at once. A rolling rate shows a small steady number and reads as healthy. The cohort view shows a cliff, and the cliff is the actual business model asking to be looked at.

Five real answers beat fifty charts

One required question on the cancellation flow, in free text. Not a multiple choice: a multiple choice can only return the reasons you already thought of, which are the ones you have already addressed. Five people telling you in their own words why they left will tell you more than any dashboard, and the sentences are usually short and unflattering.

People going quiet is next month's churn, showing early

Someone paying and not using is going to notice the charge eventually. That is the one loop where you have a window: the outcome has not happened yet, and a single message asking whether they got stuck is the whole intervention.

How this compares with everything else

There is a rough ordering of what moves a small product's revenue, and it is not the one most people work in.

Pricing comes first. It is an afternoon of work and can move revenue three to five times. It is almost always the cheapest experiment available and almost always the last one attempted.

Retention comes second, and it sets the ceiling the other two multiply against. If a fifth of your customers leave every month, acquisition is filling a bucket with a hole in it, and the size of the hole determines how much your marketing is worth.

The marketing system comes third. It is worth roughly one and a half to two times whatever the first two leave you. That is a real number and worth having, but it is a multiplier on a base the other two set.

Put concretely: at $19 a month, roughly 500 subscribers is $10k a month. Whether you ever reach 500 depends far more on how many of the ones you already have are still there next year than on how many communities you are working.

So it belongs on the same list

This is why the userbase appears as a row in the community table rather than on a separate analytics page. It has an effort figure like every other row, because retention work costs mornings like every other row. It has an MRR figure, which is usually the largest one on the screen. And it competes for the same time.

The comparison that produces is the useful one. When churn moves from 1.4% to 2.1% and nine people are at risk, that beats every new community this week. Not as a matter of taste, but because the arithmetic says so. Putting the two on the same page is what lets you see it.