Most segmentation advice is written for email marketers with a hundred thousand subscribers. You have four hundred members and you know a lot of them by name. The point of segmenting a Skool community is not sophistication — it is to stop treating four hundred people as one undifferentiated blob you address with a single weekly post.
Here is the rule I would put above everything else: a segment is only worth building if you can name the action that follows from it. "Members in Europe" is a fact. "Members in Europe who joined this month and have not posted yet" is a segment, because it comes with an obvious next move.
Which member segments are actually worth building?
The four segments worth building are lifecycle stage, tenure, paying versus free, and interest theme, because each one resolves to a list of names with an obvious next message attached.
1. Lifecycle stage
This is the highest-value cut, and it is the one that is hardest to do by eye. Everyone in your community is somewhere on a curve: brand new, warming up, cruising, cooling off, or effectively gone. The same message lands completely differently depending on where someone sits.
A workable set of stages:
- New — joined recently, has not yet formed a habit. Fragile. Deserves the most attention per head.
- Rising — activity is climbing. These people are auditioning to be your core. Notice them out loud.
- Steady — reliably present. Low maintenance, high value, easy to take for granted.
- Fading — used to be active, is measurably less active now. This is the group where a single message is worth the most, and the one covered in detail in how to spot members about to leave.
- Gone quiet — no meaningful activity for a long stretch. Some are gone. Some are one good reason away from coming back.
The crucial insight is that fading is defined against a member's own past, not against the community average. Someone who posts twice a month and always has is steady. Someone who posted twice a week for three months and now posts nothing is fading — even though both people posted zero times last week. A raw activity list cannot see that difference. Only a comparison over time can.
2. Tenure
Split by how long someone has been with you. Something like: first two weeks, two weeks to three months, and beyond three months. Tenure changes what a behaviour means. Silence in week one is a failed onboarding. Silence in month eight, after two years of contribution, is a different story and needs a different message.
Tenure also protects you from a very common mistake: sending a "we miss you" note to somebody who joined four days ago.
3. Paying versus free
If you run a paid community, or a free community with a paid tier, this cut outranks almost everything else. A paying member who is going quiet is a churn event with a date attached. A free member going quiet is a much smaller problem. When two signals conflict, buyer status should generally win.
4. Interest theme
The segment owners underuse most. Not "who is active" but "what is this person actually into". If you can group members by the topics they care about, you can post to a genuine audience rather than to an average. Ten people who care deeply about the same narrow topic will out-engage a hundred people who are vaguely interested.
Which segments look useful but aren't?
Segments built on raw post counts, likes, visits, or anything else you cannot act on feel productive and produce nothing:
- Raw post counts. A leaderboard tells you who shouts loudest. It does not tell you who is drifting, and drifting is where the money is. There is a better way to find your real core, covered in finding your most engaged members.
- Likes. A like is close to free, which makes it a weak signal of anything. Comments and poll votes cost something — attention, a bit of thought, mild social exposure. Count those.
- Anything you cannot act on. If a segment does not resolve to a list of names and one thing you will say to them, it is decoration.
- Visits. The subtle one. Segmenting on page loads gives you an "engaged" group you cannot write a message to. Skoolgrades measures a participation rate instead — comments, comment replies and poll votes — which is why its figure reads lower than a visit-based engagement rate (why that gap exists).
A good test: after you build a segment, could you write the first line of the DM without thinking? If not, the segment is not sharp enough yet.
Can I segment my members without buying a tool?
Segmenting your members needs no tool at all to start. Here is a version that works with a spreadsheet and about an hour a week.
- Export or copy your member list into a sheet. Columns: name, join date, last time you saw them do something, paid or free, and a notes field.
- Add a stage column and fill it in by eye. Be quick and rough. Rough and current beats precise and stale.
- Sort by stage. Look at the fading and new columns first. Those two are where your hour goes.
- Pick a maximum of ten people. Write to them individually.
- Next week, update the sheet before you do anything else. The update is the whole system.
This genuinely works. It falls over for two reasons, both predictable: it goes stale the moment you skip a week, and it cannot see anything that happens outside your community. That second limit matters more than people expect — see what your members are doing on Skool for why.
How does Skoolgrades segment my members for me?
Skoolgrades segments your members by reading your community through a Chrome extension running on your own logged-in Skool session. As of July 2026 there is no public Skool API — Skool ships changes often, so check its own documentation — and no scraping is involved: it is your own browser, looking at your own community, the way you already do. It reads what is publicly visible on member profiles and in public communities. It does not read your members' private DMs; chat collection is list-only, meaning the thread list and unread state, nothing inside the threads.
On top of that data, two things do the segmentation work.
Cohorts. Twenty-six saved member groups, rebuilt weekly: new joiners, at-risk, top engagers, paying members, theme groups, region buckets, members who came from ads. They refresh on their own, so the list you open on Tuesday reflects Tuesday and not the last time you remembered to maintain a spreadsheet.
The Cohort Blender. Every member gets classified on four axes, including a lifecycle state — star, rising, steady, fading, drifted, ghosted, untapped, going dark — plus how exclusive they are to your community and which activity band they sit in. That last pair is what gives you segments you could not build by hand: for instance, members who are highly active on Skool generally but almost never with you.
That specific combination has a name in the product: Untapped. The definition is deliberately concrete — at least 25 posts elsewhere on Skool, active in two or more communities, seen in the last 14 days, joined at least 30 days ago, and silent in your community for 30 days or more. These are people who obviously have the habit. They just have not built it in your room. That is a very different problem from a disengaged person, and it wants a different message.
Interest themes come from a similar place: Skoolgrades synthesises the topics a member appears to care about from their public cross-community footprint, then groups members by theme.
Building your own segment
When the pre-built cohorts do not match what is in your head, Targets lets you build a list three ways: pick a recommended group, describe who you want in plain language, or paste a post URL and pull everyone who engaged with that post. The third one is quietly the best — "everyone who commented on the post about pricing" is a razor-sharp segment that no demographic cut can produce.
From a list, you can write one message and send it either verbatim to everyone or personalised per member. You review and approve every single draft, then send them one at a time with a human click from the extension panel. Nothing sends itself. That is a deliberate design choice: blast-sending is how owners get flagged and how members learn to ignore their inbox. There is more on that tradeoff in how to DM Skool members at scale.
Which of these segment numbers should I not fully trust?
Two numbers deserve caution: cross-community counts, and segment sizes in large communities.
Cross-community counts undercount. When Skoolgrades says a member is active in four other communities, four is a floor, not a total. The collection is deliberately conservative. Treat those numbers as "at least this much", never "exactly this much".
Large communities are sampled. Above roughly 3,000 active members, collection uses a fixed sample of about 900 members rather than a full census. Segment sizes for those communities are estimates. The shape of the picture holds; the precise counts do not.
Where should I start if I have never segmented before?
The place to start is two segments rather than eight. Build two: new this month and used to be active, is not now. Write to five people in each. Do that for three weeks before you add anything more sophisticated. Segmentation only pays when it changes what you do on a Tuesday morning, and two good segments you act on beat twelve you admire.