How to vet a crypto influencer before you pay

Crypto influencers, or KOLs as the industry still insists on calling them, are the main paid channel in this market. They are also the least measurable one. You send money, a post goes out, and the numbers that come back are impressions and likes, neither of which pays your bills.

Here is the order to check things in, cheapest and fastest first. The last step is the one almost nobody runs, and it is the one that decides whether the campaign was ever going to work.

1. Engagement rate, as a filter and nothing more

Start here because it takes ten seconds and it catches the worst cases. Healthy crypto accounts on X tend to sit somewhere around 3 to 8 percent, and smaller Telegram communities run higher. Under 2 percent usually means a bought audience.

Suspiciously high is also a signal. A large account showing 15 percent and up is more likely to be in an engagement pod than to be genuinely beloved. Pods trade likes between members, which moves the ratio without moving anything real.

Treat engagement rate as a smoke alarm. It tells you something might be wrong. It never tells you the audience is good.

2. Median views, not average views

Ask for view counts and take the median across the last twenty posts. Averages get dragged upward by one viral post that has nothing to do with what you are buying.

Then compare that median against the follower count rather than looking at it alone. A 200,000-follower account with a 3,000-view median is reaching 1.5 percent of its own audience. A 30,000-follower account with a 12,000-view median is reaching 40 percent of its, and it usually costs a fraction as much.

3. Assume some of the audience is fake

This is the part people find uncomfortable, so it gets skipped.

Whether the real number is 40 or 60 in any given case matters less than the working assumption: an unscreened audience is partly fake until it proves otherwise. Bot-detection tools will give you a percentage. That percentage tells you how much of the audience to discard, and nothing about what is left.

4. Ask what the last campaign actually returned

Not impressions. Clicks, sign-ups, wallet connections, whatever the previous client was counting. A creator who has this data and shares it is already in the top decile. A creator who only has screenshots of view counts is telling you something.

If the deal is large, run a small paid test first. One post, small fee, tracked link. The test costs less than the mistake.

5. Check what the audience holds on-chain

Every step above measures attention. None of them measure money.

An audience can be entirely human, engaged, awake, and still hold nothing. That audience will like your post and will not buy anything, because it cannot. In crypto you can check this directly, which is unusual and mostly unused.

Resolve the creator's followers to their on-chain wallets and look at four numbers:

• How many followers resolve to a wallet at all. Bots and ghost accounts have no on-chain history, so this doubles as a fake-follower check that no one can game cheaply.
• Total value held across the resolved wallets. The ceiling on what this audience could ever spend.
• Median balance. The honest number, because one whale distorts the total and the median does not care.
• The distribution across balance tiers. Ten wallets over $100k is a very different audience from four thousand wallets under a dollar, even when the totals match.

This reorders your shortlist more often than people expect. The account with the best engagement rate frequently has the poorest audience, because cheap engagement is easy to accumulate and funded wallets are not.

What the cost of skipping this looks like

Note what happened there. The fake followers did not merely waste part of the budget. They tripled the real cost of every user the campaign produced, because the spend stayed constant while the denominator collapsed.

The short version

1. 1Engagement rate under 2 percent or over 15 percent, walk away.
2. 2Median views against follower count, not average views.
3. 3Assume a large share of the audience is fake and price accordingly.
4. 4Ask for outcomes from the last campaign, and test small first.
5. 5Resolve the audience to wallets and look at the median, not the total.

Steps one through four are standard practice and most teams do at least some of them. Step five is the one that separates an audience that likes your post from an audience that can buy your product.