2026-08-23 · 5 min read
Follower activity cannot be measured — but one number can
The tools claiming to measure follower activity are measuring three different things and reporting one conclusion. Pull them apart and only one comparable metric survives.
An account with 100k followers posts something. Two hundred views. Someone replies: your followers must all be bots.
They might be right. But they cannot prove it — "follower activity" still has no definition anyone can verify.
The short version
- It is really three things, and they multiply
- Any one of them collapsing produces the same ugly number — so a single post proves nothing about follower quality
- The only comparable metric: how many distinct people you moved, counted by one published rule
Three things on one chain
1. Authenticity — whether the followers are real people. Bought accounts, bots, people who last logged in three years ago.
2. Delivery — whether the algorithm puts your post in front of them. This has nothing to do with whether they are real. Two months of silence, an external link, a run of weak engagement: distribution drops.
3. Response — it arrived; how many actually did something.
So "100k followers, 200 views" is equally consistent with three different diseases: 100k bots; 100k real people the algorithm never reached; delivered fine but nobody wanted it. Three causes, one reading.
It runs the other way too. A 10k account that suddenly gets 50k views may have ridden a trending term, with 45k of those views coming from strangers in the recommendation feed — the follower layer did almost none of the work.
Inferring follower quality from a single post is the wrong direction entirely.
Four common methods and what breaks in each
| Method | What it measures | The problem |
|---|---|---|
| Engagement rate = (likes + reposts) / followers | A proxy for layer 3 | Likes are the cheapest action and the easiest to buy — and the people liking need not be your followers |
| Third-party bot checkers | The surface of layer 1 | Judged on avatar, post count, follow ratio. A real person dormant for three years scores as a quality follower |
| Native impressions | Layer 2 × layer 3 | Closest to the truth, but bundles non-follower reach with no way to separate it, and only the owner sees it |
| Native link clicks | Closest to behaviour | Not deduplicated; includes crawlers and link previews; one person refreshing three times counts three times |
That last row deserves a sentence more: a like is an opinion, opening a link is a behaviour. It costs attention, which makes it the highest-quality signal on the list — but the platform's version of the number is too dirty to use and visible only to the owner. Two people quoting private numbers are not comparing anything.
So what should be measured
Replace the question:
When I post something, how many distinct people actually move somewhere else?
- It measures behaviour, not sentiment. A like is reflexive; leaving the platform is not.
- It can be deduplicated, and the rule can be published. One person clicking three times counts once.
- It can be compared across accounts — given the same rule, the same page, the same window.
The third is the one no native dashboard offers: numbers produced by different rules are not comparable at all.
What it still cannot tell you
It does not measure follower activity. It measures the effect of one post. How good the idea was, how strong the hook was, when it went out, whether the algorithm felt like distributing it — all baked into the same number, none of it separable. Any product claiming to isolate follower activity is dodging this.
So one result means nothing; a run of them means something. A single post can be carried by one good idea. The median of twenty cannot.
Measure it yourself: four steps
- Use a tagged link with any analytics tool (Plausible, GA) and read unique visitors, not pageviews.
- At least five runs, take the median, not the mean. One outlier makes an average meaningless.
- Post at a fixed time. One less variable, one more credible conclusion.
- Record impressions too, and compute clicks ÷ impressions — not clicks ÷ followers.
Step four is the most useful thing here, because it is what splits layer 2 from layer 3:
| What you see | What it means | What to fix |
|---|---|---|
| clicks ÷ followers is low | Nothing at all | — |
| impressions ÷ followers is low | Never delivered (layer 2) | Posting cadence, account standing |
| clicks ÷ impressions is low | Delivered, unwanted (layer 3) | The idea and the hook |
Most people, before splitting those two ratios, are busy fixing the wrong one.
Why I built this
clicksnotfollowers exists for comparability: everyone measured by the same dedup rule, results put side by side in public.
It cannot measure follower activity either. It measures exactly one thing: you post once, and this many real people actually clicked.