CBCreatorBenchmarks
JournalMethodology9 min read

The 2026 fake-follower study,
and why we won't give you a percentage.

We read 41,000 public profiles. Here's what authenticity signals can and can't tell you about audience quality.

Every authenticity tool on the market gives you a percentage. “42% fake followers.” “83% real audience.” The number looks like data. It is not.

Across forty-one thousand public Instagram and TikTok profiles, we ran every signal you can read without the platform’s private API: like-to-comment ratio, comment depth, comment emoji density, engagement variance across recent posts, follower-to-following ratio, account-age distribution. We compared the resulting composite to the “fake-follower percentages” from three named competitors. The two correlate at r = 0.31.

A correlation of 0.31 means the ‘percentage’ is roughly nine percent better than a coin flip.

What we can actually tell

Public engagement signals do reveal a lot — they just don’t reveal a percentage. They reveal whether the engagement on a profile looks like the engagement of similar profiles in the same cohort. That’s a useful sanity check. It is not a fraud verdict.

  • Like-to-comment ratios above 200:1 are unusual outside the meme niche.
  • Comment depth below 8 characters average is unusual at scale.
  • Engagement variance below 12% across the last 30 posts is unusual for organic accounts.
  • Follower-to-following ratios above 50:1 with recent rapid growth are unusual.

Why we say it the way we say it

On a CreatorBenchmarks report, the authenticity section reads things like “sits within the typical range” or “outside the typical range.” That’s the most honest thing the data supports. A percentage isn’t.

We could give you a percentage. The button is one ALTER TABLE away. We won’t.