DEV Community

Oznaka
Oznaka

Posted on Originally published at oznaka.ai

Spotting Fake Influencers: What We Check Before Paying a Crypto Creator

The fake influencers that cost us the most never looked fake. An account with 80,000 followers and 15 likes a post gets caught in seconds. The expensive ones had a tidy feed, a believable audience and a price that matched their size. The problems only showed up after the post went live.

After three years of running KOL campaigns, we stopped asking whether an account is fake. We ask a narrower question: if we pay this creator, how many real people in our market will see the post and care?

What gets faked

Usually one of four things, sometimes all of them:

  • the audience: bought followers or giveaway hunters;
  • the engagement: bot replies, pods and raid groups that trade replies;
  • the reach: paid boosts shown to clients as organic views;
  • the opinion: sponsored posts with no ad label.

Red flags you can check by hand

No single sign proves fraud. Three or more on one account means slow down.

  1. Follower spikes with no viral post, repost or interview behind them.
  2. An engagement rate far from creators of the same size in the same niche, in either direction. Take (likes + replies + reposts) ÷ followers over the last 10–20 posts.
  3. Replies that would fit under any tweet: “LFG”, “bullish”, AI summaries of the post.
  4. The same 30–40 accounts replying to every post within minutes.
  5. Views with almost no likes, or likes with almost no replies. Real replies are the hardest thing to fake at scale.
  6. A new “token I've been researching for weeks” every few days, with no ad label.
  7. One huge post among many small ones, often a past sponsored post. Price creators on their median post.

The five-minute routine on X

Open “About this account”: since November 2025, X shows the country or region, the join date and how many times the username changed. Scroll 20 posts and count sponsored posts, giveaways and original thinking. Open five reply profiles under three posts; this step alone catches most engagement farming. Then search the handle together with projects they promoted, and look for quote posts that now say “This post is unavailable.”

Why it's worth the time

A Harvard Business School study of 36,000 tweets from 180 crypto influencers found that promoted tokens fell 6.53% on average within 30 days and 18.9% within 90. In Influencer Marketing Hub's 2026 benchmark, fake or bot followers made up 56.5% of reported fraud and quality issues.

Put a number on it

Cost per real reader = price ÷ (median views × organic reach share × niche match share)

Two creators of the same size with the same quote can differ twofold on this number. That gap is what a fake audience really costs.

Manual checks work for five creators, not fifty, and a profile never shows how much reach was boosted. On X, that part is what Oznaka (oznaka.ai) scans.

Full guide, with Instagram and YouTube checks: https://oznaka.ai/guides/how-to-spot-fake-influencers

Top comments (0)