The numbers flag, a person decides
- In August 2026 we tested our own three-layer check on 9 athletes matched to an energy drink brand.
- Our first screen, built on the engagement numbers we store, flagged 5 of the 9.
- A person then opened each flagged account. Only 1 flag held up.
| Flagged account | What a person found |
|---|---|
| First | 9 of 12 posts hid their likes, and our stored average counted each hidden one as minus 1 |
| Second | Weak on likes, but 290 typical comments, the second most in the group |
| Third | Held up: 16,288 typical likes against 38 typical comments |
| Fourth | The account keeps only 3 posts |
| Fifth | No engagement data on file at all |
Source: Our own vetting run on 9 athletes, 2026-08-09
Judge the typical post
- Demand the median, not the best post.
- In the same run, one account had a single post at 380,722 likes, against a typical post of 13,000 to 35,000. Its average made it look far bigger than it is.
- Comments still show when likes are hidden, so the typical comment count was the one number we could compare across all 9.
Check every brand that has paid them
- Every one of the 8 new names had a past drinks deal on record. A plain "no competing products" rule left us zero names.
- So we split that rule three ways:
- Same shelf: a drink a shopper could buy instead of yours.
- Wrong message: a deal that points their audience the other way, like a spirits brand on a fitness creator.
- Live exclusive: the only one that legally stops you. None of the 8 had one, and every deal on record was from before 2024.
- Then read the source behind each deal. Our data called 3 of the 8 spirits endorsers. A person read it: they had appeared in a spirits brand's interview series, which is a campaign appearance.
Then read what they've said
- No number reads old posts. A person goes through their videos, captions and comments, going back years, and every brand they've promoted.
- That's the part a fake follower check skips, and our guide on those tools covers what each one reads.