Every community owner hits this wall. You post something you were sure would land. Three views, no comments. Then you dash off a throwaway question on a Sunday night and it runs to forty replies. Whatever model of your members you had in your head is clearly wrong, and the obvious fix — ask them — turns out to be less obvious than it looks.
Why can I not just ask my members what they want?
Surveys mislead because they capture what members say they want rather than what they actually show up for, and they answer a narrower question than owners think. When you ask members what they want, you get one of three things:
- The aspirational answer. People describe the version of themselves they want to be. They will tell you they want advanced technical deep-dives, then engage with the beginner thread.
- The polite answer. You are the owner. Members like you. They will nominate whatever they think you want to hear.
- The last-thing-that-annoyed-them answer. Perfectly real, but it is feedback on a moment, not a signal about direction.
None of that is dishonest. It is just that stated preference and revealed preference are different things, and only one of them predicts what people will actually show up for.
The useful version of asking is not a survey. It is ten one-to-one conversations where you shut up and let them talk about their week.
So how do I find out what my members really want?
Three sources tell you what members really want: what already earns comments in your own room, what they engage with elsewhere on Skool, and ten one-to-one conversations.
1. What already gets engagement in your own room
Your own community is the highest-quality signal you own, and most owners barely read it. Go back over the last two or three months of posts and mark each one with a topic tag and its comment count. Not likes — comments and poll votes. A like costs a member nothing, which makes it a weak indicator of anything. A comment costs attention and a small amount of social exposure. Those are the units worth counting.
Two things to control for while you do it:
- Timing. A great post published when nobody is around underperforms a mediocre one published at the right hour. Do not conclude the topic was wrong when the hour was.
- Maturity. Compare posts that have had a few days to breathe. Fresh posts always look weaker than they will end up.
You are looking for the clusters. Usually three or four topics carry most of the conversation, and one of them will surprise you. If you want the format side of this rather than the topic side, what kind of Skool post gets comments covers it.
2. What members care about outside your community
This is the source almost nobody uses, and it is the strongest one available on Skool specifically. Skool is a network. Your members are in other communities — some in your niche, some nowhere near it — and what they choose to spend attention on there is preference expressed with no social pressure from you at all.
If eleven of your members are all deeply engaged in rooms about a topic you have never covered, that is not a coincidence, and it is not something a survey would have surfaced. They would not have thought to mention it. You would not have thought to ask.
This also fixes the misread that costs owners the most. A member who is silent with you but posting daily elsewhere is not disengaged. They are engaged, just not here. That is a positioning problem, not a retention problem, and the fix is completely different. There is more on that in where else your members are active on Skool.
3. Ten conversations, not one survey
Pick ten members across different stages — two new, two who are clearly thriving, two who used to be active and have gone quiet, four at random. Message each of them individually. Do not ask "what do you want from the community". Ask what they are working on this month and what is currently in the way.
This pairs well with a general read of what your members are doing on Skool, so you know who to pick. You will get more usable direction from those ten replies than from two hundred survey responses, because you can hear the hesitation and ask the follow-up.
How does Skoolgrades work this out?
Skoolgrades collects your community through a Chrome extension running on your own logged-in Skool session — no Skool API, no scraping, no infrastructure sitting between you and Skool. To Skool it looks like an admin browsing their own community, because that is exactly what it is. It reads what is already publicly visible: profiles, public posts, public communities. It does not read your members' private DMs. Chat collection is list-only, meaning the thread list and unread state and nothing else.
From that, two pieces speak directly to this question.
Member interest themes. Skoolgrades synthesises, per member, the topics that person appears to care about, drawn from their public footprint across Skool. Grouped up, that gives you a picture of what your audience cares about that is built from behaviour rather than from what anyone told you. It also gives you theme-based member groups, so "everyone whose interests cluster around this topic" is an actual list of names rather than a hunch.
Attention Pull. This ranks external Skool rooms by the share of your members' attention they capture, not by raw headcount. The share framing matters. Ranking by raw numbers just surfaces whichever huge community happens to contain a few of your people. Ranking by share surfaces the rooms that own a real slice of many members' attention — which is a much better description of where your audience actually lives.
Alongside that, Skoolgrades maintains twenty-six cohorts that rebuild weekly, including theme groups, and runs a sentiment and community-themes read over your own content so you can see what topics dominate inside your room and what the general mood is.
How do I turn a signal into an actual post?
Turning a signal into a post takes five steps, because a theme on its own is not an action:
- Pick one theme that shows up in both places — your own engagement history and members' outside interests. Overlap is the strongest evidence you will get.
- Write one post about it, in the most specific form you can manage. Specific beats broad every time. "How I priced my first cohort" beats "thoughts on pricing". There is a fuller treatment in what to post in your Skool community.
- Publish it when your people are around. Timing readouts are worth checking before you conclude a topic failed.
- Tell the right people it exists. Build a list of members whose interests match the theme, write one message, and send it. In Skoolgrades that means Targets plus DM Broadcast: you build the list, choose verbatim or personalised-per-member drafting, review and approve every draft, and then send them one at a time with a human click from the extension panel. Nothing sends itself, ever.
- Read the result honestly, after a few days, on comments and poll votes.
Run that loop four or five times and you will have a real model of your members instead of a guess.
What are the limits of this data?
Two limits are worth stating plainly before you lean on these numbers.
Cross-community counts undercount on purpose. If Skoolgrades tells you nine members are active in a given room, treat nine as a floor rather than a total. The collection is deliberately conservative, and a number you can trust as a minimum is more useful than one you have to caveat.
Large communities are sampled. Above roughly 3,000 active members, collection uses a fixed frame of about 900 members rather than a full census. Theme distributions for those communities are estimates. The ranking is generally stable; the exact percentages are not.
What can the data not decide for me?
Data will tell you which topics have pull, but it will not tell you which of those topics you should be the person to teach. That judgement stays with you, and it should — a community that chases whatever its analytics point at ends up with no spine.
Use behaviour to rule things out and to find the two or three themes you were blind to. Then decide, as the owner, which of those you actually want to build around. The point of reading your members well is not to hand them a menu of what they asked for. It is to stop guessing about the things you could simply have looked up, so you spend your judgement where judgement is the only thing that works.