Bought followers never bring likes with them. That single fact is the whole test: the purchase inflates the denominator of the engagement rate and leaves the numerator exactly where it was. So the account ends up with the follower count of one tier and the like count of a much smaller one — and that mismatch is measurable, unlike "their followers look like bots."
The test: followers say one tier, likes say another
Here's what a real post actually collects at each size, across 17,000+ real creator profiles:
| Follower tier | Median likes | Median comments | Median ER |
|---|---|---|---|
| Under 10K | 125 | 9 | 2.23% |
| 10K–50K | 292 | 17 | 1.52% |
| 50K–500K | 2,616 | 57 | 1.93% |
| 500K+ | 31,565 | 247 | 2.46% |
Read the account's follower count, then read its likes, then check whether both point at the same row. An account with 180,000 followers pulling 300 likes a post is claiming the third row and performing the second. That's close to a ninefold gap between the audience it sells and the audience that shows up.
The gap also tells you how much was bought, approximately. Likes of 300 belong to a creator in the 10K–50K band. So the real audience behind that 180K account is plausibly somewhere in the tens of thousands, and most of the rest is dead weight.
Why "low engagement rate" alone gets this wrong
Nearly every guide on this says the same thing: low engagement relative to followers means bought followers. The problem is that engagement rate doesn't fall in a straight line as accounts grow. In our data it dips and then climbs back — 2.23% under 10K, down to 1.52% at 10K–50K, back up to 1.93%, then 2.46% at 500K+.
So the 10K–50K band is naturally the weakest-engaging tier we measure. It's also where follower-buying is most common, because that's the range where a creator is trying to look pitchable to brands. A soft engagement rate there is the least diagnostic signal available — it's what honest accounts in that tier look like too.
Niche moves the floor even further. Education creators sit at a 0.66% median and communities at 0.97%. A blanket "under 1% means fake" rule accuses two entire legitimate categories.
The honest suspicion threshold
If you want one number per tier, use the 25th percentile — the line three of every four real creators clear:
| Tier | Bottom-quartile line |
|---|---|
| Under 10K | 1.07% |
| 10K–50K | 0.61% |
| 50K–500K | 0.69% |
| 500K+ | 0.94% |
Below that, the account is in the slowest quarter of real creators its size. That's not proof of a purchase — plenty of genuine accounts live there — but it's the point where the like-count check above is worth running.
Bought likes look like the opposite problem
"Bought followers" and "bought likes" are separate purchases with opposite signatures, which is why one test can't catch both. Followers drag the rate down. Likes shove it up, often past the 5.56% top quartile, and they arrive without comments — real audiences produce both together, as the pairs in the first table show. If the likes are impossible and the comments are thin or generic, you're looking at the inflation problem instead; that's covered in spotting fake Instagram engagement.
The signs worth checking by eye
Only three of the usual eyeball tests actually earn their place:
- A step, not a curve, in follower growth. Real growth is ragged. A flat line with a vertical jump is a transaction.
- Follower geography that doesn't match the content. An account posting in English to an audience concentrated in countries where nothing they post lands is the cheapest tell there is.
- Comment authors who never reappear. Real audiences have regulars. Bought engagement has strangers, once each.
Skip the rest. "Followers with no profile picture" was a useful signal years ago; the farms fixed it.
Can you actually prove it?
No — and anyone selling certainty here is overselling. Instagram doesn't publish audience quality, and follower purges muddy the record further by shrinking a follower count while old bought likes stay put. What you can do is measure the distance between what the account claims and what its posts deliver, then decide whether that distance is worth your budget.
Run the handle through the engagement checker to get its real rate from actual recent posts, then hold it against its own tier and niche on the benchmarks. An account that misses its tier's like count by an order of magnitude, post after post, has already answered the question.