On this page
- The wrong room is a measurement problem before it is a content problem
- The Ra-Aha Five-Bucket Census, and the rules that keep it honest
- Running the census takes about two hours, and here is the sequence
- What your Buyer Share has to be, worked backwards from your own pipeline
- Reading the result: five skews and what actually causes each one
- The wrong room gets wronger without any help from you
- Four levers move composition, and they work at very different speeds
- Sometimes the peer-heavy room is the correct answer
- Re-measure at sixty days, and know what counts as movement
The wrong audience is a measurable fact, not a vague feeling. Take your last twenty posts, list everyone who reacted or commented, and sort each name into five buckets using their headline: buyer, gatekeeper, peer, aspirant, ambient. The share falling in the first two buckets is your Buyer Share. That single number tells you whether you have a content problem or a distribution problem, and posting more only helps with one of them.
The wrong room is a measurement problem before it is a content problem
You can settle this question in an afternoon, and almost nobody does. The complaint is always phrased as a feeling. My posts do well and none of the people cheering can hire me. Everyone who engages is another consultant. The response the internet gives back is to post better content, or to be more specific about your niche, and neither of those instructions can be checked. Two months later you still do not know whether anything moved.
Audience composition is a count. Every person who reacted or commented on your posts has a headline attached to their name, and a headline is a rough but usable statement of what somebody does for a living. If you read four hundred of those headlines and sort them against your own definition of a buyer, you get a percentage. The percentage is either acceptable or it is not, and in sixty days you can compute it again and see whether the work you did had any effect.
That is the whole gap in the published advice on this topic. One or two pages will tell you that most creators attract peers rather than buyers. None of them hands you the method that produces your own figure, which means the reader ends up with a diagnosis and no instrument.
of members post more than once a week. The visibly active layer of the platform is small, which matters here for one specific reason: your engager list is not a sample of your audience. It is a sample of the habitually active, and habitually active people are disproportionately people who are building a presence of their own.
Aggregate 2026 LinkedIn statistics reportsThe Ra-Aha Five-Bucket Census, and the rules that keep it honest
Counting is the easy half. The hard half is deciding which bucket a headline belongs in, which is where most attempts at this quietly collapse into wishful thinking. Five buckets, one bucket per person, and a tie-break rule that stops you from promoting people into the buyer column because it feels better.
Running the census takes about two hours, and here is the sequence
Twenty posts is the working window. Fewer than that and one unusual post dominates the result. More than that and you are measuring a version of yourself who has already changed. The steps below assume you are doing this by hand, because the by-hand version works on every account regardless of what your analytics screen happens to offer you this month.
Take your last twenty posts. Find the median engagement count, then set aside any post that earned more than three times it. Those posts reached a different room and belong in their own separate count, because a single post that travelled will otherwise decide your entire percentage. Audit the outliers afterwards as a second, smaller census, and the comparison between the two is usually the most interesting output of the whole exercise.
Open each post's reaction list and its comment thread, and copy two fields per person into a spreadsheet: name and headline. Whatever aggregate demographic breakdown your account gives you is a useful cross-check and a poor substitute, because those title groupings are the platform's taxonomy of jobs and not your definition of a buyer.
Somebody who reacted to fourteen of your twenty posts is one person. Counting them fourteen times is precisely how a small loyal group of peers comes to look like an audience. Collapse the list to unique people and keep a frequency column beside each name, because you are about to need both numbers.
Ten seconds per headline, one bucket each, no revisiting. Classifying across several days introduces drift, because your standard for what counts as a buyer moves as you get tired. If you cannot decide within ten seconds, that hesitation is itself the answer and the person goes to ambient.
Buyer Share by unique person, which is buyers plus gatekeepers divided by unique engagers. Then Buyer Share by engagement event, which uses the frequency column as a weight. Report both. The gap between them is a separate diagnosis, covered further down.
The time cost, with every input labelled as an assumption you should replace with your own: twenty posts at a median of thirty reactions and six comments gives you 720 engagement events, which usually deduplicates to somewhere around 400 unique people. Collection runs about forty minutes. Classification at ten seconds a headline is 400 times ten, which is sixty-seven minutes. Call it two hours including the spreadsheet work.
What your Buyer Share has to be, worked backwards from your own pipeline
There is no published benchmark for this and anyone quoting you one has invented it. The useful target is derived rather than looked up, and it comes from the only question that matters: how many buyer-side humans do you need in front of you each month for the rest of the funnel to work? Work backwards from that and the required percentage falls out.
| Input or result | Scenario A | Scenario B | Scenario C |
|---|---|---|---|
| Monthly engagement events (assumption) | 600 | 600 | 600 |
| Unique engagers after deduplication (assumption) | 320 | 320 | 320 |
| Buyer Share by unique person | 4% | 12% | 25% |
| Buyer-side people reached per month | 13 | 38 | 80 |
| Buyer-side people reached per year | 156 | 456 | 960 |
Illustrative arithmetic. Every input is an assumption. Replace the top two rows with your own figures and the rest recalculates.
Whatever your conversion rate is from a buyer-side engager to an actual conversation, it is the same rate in all three columns. So moving from scenario A to scenario B does exactly what tripling your posting volume would do, and this is where the comparison becomes uncomfortable for the standard advice.
Going from three posts a week to nine, at one hour per post including thinking time, is six extra hours a week. Across forty-eight working weeks that is 288 hours a year. The census plus one deliberate round of subject changes is closer to eight hours. Both routes get you to roughly the same place in the table above. One of them costs 280 hours more than the other, and it is the one that gets recommended.
The engagement volumes, the deduplication ratio and the three Buyer Share values are illustrative. What is not assumed is the structure. Buyer-side reach genuinely is the product of total reach and composition, which means composition is a multiplier on every impression you will ever earn rather than an addition to this month's.
- Advice to post better content is unfalsifiable, because nothing about it can be checked against your own audience two months later.
- Two hours with a spreadsheet turns the complaint into a percentage: your Buyer Share, computed from the engagers on your last twenty posts.
- Compute the number twice, once per unique person and once per engagement event, because the gap between the two is a different problem with a different fix.
- Peers and aspirants engage more per impression than buyers do, so a distribution system that rewards engagement will quietly steer you further away from the people who buy.
- Tripling your Buyer Share and tripling your posting volume produce the same arithmetic result, and one of them costs roughly 280 fewer hours a year.
Reading the result: five skews and what actually causes each one
A Buyer Share on its own tells you the size of the problem. The shape of the distribution across the other buckets tells you the cause, and each cause has a different first move. Find your row before you change anything, because the four available levers work on completely different timescales.
| What the census shows | Most likely cause | What to change first |
|---|---|---|
| Peers above 40%, buyers in single digits | You are writing about your craft, in the vocabulary of your craft, which only your craft can appreciate | Change the subject from how you work to the decision your buyer is making this quarter |
| Aspirants above 30% | Your posts teach people how to do your job, which is the most shareable thing you can write and the least commercial | Move from how-to toward how-to-decide, and from tactics toward tradeoffs and their costs |
| Ambient above 50% with buyers scattered thinly | Your headline is sorting the wrong people into the room before they ever read a word | Fix the sorting mechanism, starting with the headline that sorts visitors |
| Event-weighted share far below person-weighted share | A small loyal group produces most of your visible engagement while buyers read once and leave | This is distribution rather than subject matter. Go to where buyers already gather, using commenting into buyer-side rooms |
| Buyer Share healthy, conversations still zero | Composition is fine and conversion is broken, which is a completely different repair | Stop editing your content. The leak is on the profile or in the opening message |
| Fewer than 50 unique engagers in the window | The sample is too small to classify and the percentage is noise | Raise volume first, re-run the census once the window contains 50 or more unique people |
Read your census against this table before choosing a lever.
The fifth row is the one people resist hardest, because it means the content was never the problem. If eighty buyer-side people encountered you last month and none of them wrote to you, adding a ninth post a week is an expensive way of avoiding a much shorter conversation with yourself about what your profile and your first message actually say.
The wrong room gets wronger without any help from you
Composition drifts in one direction, and the mechanism is straightforward once you see it. Distribution on any feed is decided partly by early engagement. The people most likely to engage early are the people who see you most often, which is the people who already engage with you. That loop alone would preserve whatever composition you started with. What tilts it is that the buckets do not engage at equal rates.
A peer has a professional reason to be in the comments, because visible engagement is part of how they build their own presence. An aspirant has an educational reason, because your post is their training material. A buyer has neither. A buyer reads the post, forms a quiet opinion about whether you know what you are talking about, and goes back to work without touching anything. The most valuable reader you have is the one who leaves the least evidence.
So per impression, a peer-heavy audience produces a higher engagement rate than a buyer-heavy one. Any system that reads engagement as a proxy for quality will therefore show more of the peer-pleasing version of you to more people who resemble peers. Your engagement rate rises. Your Buyer Share falls. Both things are happening at once and only one of them is on your dashboard.
That is a mechanism argument, not a published measurement, and no platform releases engagement rates broken down by the commercial relationship between reader and author. The reason to trust it is that you can test it on your own data in two hours using the census above, which is exactly the point of computing your own figure rather than borrowing someone else's.
Four levers move composition, and they work at very different speeds
Composition responds to four things, and only four. Pick one per cycle. Changing three at once guarantees that whatever happens next teaches you nothing, because you will have no way of attributing the movement.
| Lever | What it actually changes | How fast composition responds | Cost to you |
|---|---|---|---|
| Subject swap | Which profession finds the post relevant enough to stop scrolling | Two to four weeks | Free, and genuinely uncomfortable, because your existing audience liked the old subject |
| Distribution swap | Whose comment sections you appear in, which decides who sees your name before they see your posts | Three to six weeks | Several hours a week, permanently |
| Invitation policy | The standing composition of your follower base, which is the prior on every post you publish | Eight to twelve weeks, the slowest and the most durable | Minutes a week |
| Headline and profile sorting | Who converts from a visit into a follow, which is upstream of everything else | Immediate for new visitors, no effect on people already following you | One afternoon, once |
The four levers, with honest timescales. Speeds are judgements about mechanism, not measured averages.
The invitation policy deserves a sentence of its own because it is the lever people leave switched off. Accepting invitations is close to free and you can accept everyone. Sending them is the controllable half, and every invitation you send to a peer is a small permanent vote for the composition you say you want to change. If you send ten invitations a week and all ten go to people who do what you do, no amount of subject swapping will outrun it.
One further note on the subject swap, since it is the lever most likely to be misread. The instruction is not to write better. It is to write about a different thing. A post about how you structure a discovery call is craft, and craft attracts practitioners. A post about how a buyer should decide whether to run the project internally is a decision, and decisions attract the people who have to make them.
Sometimes the peer-heavy room is the correct answer
Before you spend a quarter fixing your Buyer Share, check whether the census actually told you something bad. Three situations turn a peer-heavy audience from a failure into the intended outcome, and in all three the mistake was in the bucket definitions rather than in the content.
- Your revenue arrives through referral, subcontracting, partnerships or speaking. In that model peers are buyers, and you should have put them in bucket one before you started counting.
- You are hiring. Aspirants are a recruiting pipeline, and a strong aspirant share is worth real money in reduced agency fees, just not on the revenue line.
- You sell into a market where practitioners recommend and executives sign. The practitioner belongs in the gatekeeper bucket, and a healthy gatekeeper share is exactly what the early stage of that motion looks like.
This is also the reason to set a floor rather than a target. A room with no peers in it has no referrers, no reshares from people with adjacent audiences, and no one who understands your work well enough to argue with it in public. Deciding that a Buyer Share below some number is unacceptable is a useful commitment. Chasing a Buyer Share of 100% would strip out the part of your audience that carries you into rooms you cannot reach yourself, which is precisely the value of commenting into buyer-side rooms working in the other direction.
Re-measure at sixty days, and know what counts as movement
Sixty days is the earliest honest re-read, and the second census has to be run the same way as the first or the comparison is worthless. Same window length, same rubric, same person classifying, same time of day if you can manage it. Then apply a noise floor before you celebrate anything.
- Same twenty-post window length, with outliers quarantined by the same three-times-median rule.
- Same written ICP sentence, unedited. If you changed the definition, you changed the instrument and the comparison is void.
- Noise floor applied: on a 400 person sample, one person is 0.25 percentage points, so a three point move is twelve people and one ambiguous judgement call can produce it. Treat anything under five points as noise.
- Both numbers recomputed, per person and per event, because the two can move in opposite directions and that gap is the finding.
- New followers checked separately against the target titles, since followers are the slower and more durable signal.
- Exactly one lever changed during the period, written down at the start so you cannot retrofit an explanation afterwards.
If composition has moved and conversations have not, the census has done its job and handed the problem to the next stage, which is what your profile does with a buyer who arrives, and what you say when you write to one of them directly. That is a different discipline with a different failure mode, covered in direct messages to named executives.
Post better content is not advice, because there is no version of the next sixty days in which it turns out to have been wrong.The test any diagnosis has to pass
Questions people ask next
How many posts should the census cover?
What if I cannot see who reacted, only a total count?
Should I classify followers or profile viewers instead of engagers?
Is a large audience of peers worth anything at all?
How do I classify a headline that says something vague like founder?
Will changing my subject matter cost me the audience I already have?
Send your profile, keep the aha.
One email with your LinkedIn URL. One specific, actionable insight back, free. If it is obvious, we did not look hard enough.
Email your profile →
