Start with the follower base
The first pass takes two minutes and removes the obvious cases. Open the account's followers list and read the first fifty. On a real crypto account you will see a mix of named accounts, other traders, a few projects and some anonymous profile pictures. On a bought account you will see rows of accounts created in the same month, with default avatars, no posts and follower counts under ten.
Then look at the growth pattern. Tools that chart follower history will show you whether the account added 40,000 followers in one week in a month when they posted nothing special. Real growth from a viral post shows up as a spike followed by a plateau. Bought growth shows up as a vertical line, or as a perfectly smooth ramp that never dips, because real accounts always lose some followers every day.
One more quick check: compare followers to following. An account with 300,000 followers that follows 290,000 people bought its audience through follow-for-follow. That audience does not read anything.
Do the engagement rate math
Engagement rate is the number that makes accounts comparable across tiers. Calculate it over the last ten non-promotional posts, not the best one.
Engagement rate = (likes + replies + reposts + quotes) divided by followers, times 100.
Take the average across those ten posts. Compare it to the band for the account's size, because big accounts always have lower rates than small ones and that is normal, not a problem.
| Followers | Weak | Healthy | Suspicious, check comments |
|---|---|---|---|
| Under 10,000 | Below 1.5% | 2% to 6% | Above 8% |
| 10,000 to 50,000 | Below 0.8% | 1% to 3% | Above 5% |
| 50,000 to 250,000 | Below 0.4% | 0.6% to 2% | Above 3% |
| 250,000 to 1,000,000 | Below 0.2% | 0.3% to 1% | Above 2% |
| Above 1,000,000 | Below 0.1% | 0.15% to 0.6% | Above 1.5% |
A rate above the band is as much of a warning as one below it, because the cheapest way to fake quality is to buy engagement on a small number of posts. The engagement rate calculator does the arithmetic and compares the result to these bands for you.
Also check the ratio between reply count and like count. Real crypto accounts get argued with. If an account has 3,000 likes and 11 replies, the likes are probably not from people.
Read the comments
This is the check that cannot be faked cheaply and it is the one people skip. Open the last five posts and read the first thirty replies on each.
- Good signs: replies that disagree, replies that reference something specific in the post, questions about details, people quoting the post with their own take, the same recognisable handles showing up across weeks.
- Bad signs: strings of emoji, one-word replies like great or bullish, replies in a language the account does not post in, accounts replying within seconds of every post, and dozens of near-identical compliments.
- Worst sign: replies that are all from accounts who also reply to the same five other KOLs within minutes. That is an engagement pod, and the members are paying each other rather than reading.
If you would not want your project discussed by the people in those replies, do not buy the post.
Check where the audience actually lives
An account can be completely real and still useless to you because its audience is in a market you do not sell to, or one you cannot legally sell to. There is no perfect public tool for this, so triangulate.
Look at the language of the replies, the time of day when engagement peaks, and the location fields on the fifty most active repliers. If the peak engagement is at 3am in the KOL's stated time zone, the audience is somewhere else. Ask the KOL for the audience breakdown from their own analytics, which on X sits under the account analytics tab and on YouTube under the audience tab. Ask for a screen recording rather than a screenshot.
This matters most for compliance. If a large share of the audience is in a jurisdiction where your token cannot be promoted, that reach is a liability rather than an asset.
Look up what they promoted before
Search their account for the words they use when they are paid. On X, search from:handle with terms like partner, sponsored, ad, CA, or the word gem. Then take every ticker you find and look up what happened to it.
You are looking for a pattern, not a single bad call. Everyone who has been in crypto for five years has promoted something that failed. What should stop you is an account that promoted six tokens in a year that all peaked within 48 hours of the post and then fell 95%. That is not bad luck, that is the business model, and their audience has learned it.
Check for deleted posts too. Search the account on the Wayback Machine and compare. An account that deletes its promotions after the token dies will delete yours as well, which means you paid for something that no longer exists. Put a no-deletion clause in the contract, as covered in KOL contract terms.
Wallet checks
If a KOL has a public wallet or a known ENS name, you can see what they actually do. Look at whether they bought the tokens they promoted, whether they sold within hours of posting, and whether they received tokens from a project treasury before the post went out.
Be honest about the limits here. Most KOLs use more than one wallet, a wallet you find is not proof of anything on its own, and being unable to find a wallet says nothing at all. Treat this as a check that can disqualify someone rather than one that can approve them. If you find a wallet that received an allocation and dumped it into the post's own volume, that is enough to walk away. If you find nothing, move on and rely on the other checks.
Tools that help
| Tool | What it is for |
|---|---|
| Native X and YouTube analytics | The only trustworthy source for audience geography and real views. Ask for a screen recording. |
| Follower audit tools such as HypeAuditor or Modash | Fake follower estimates and audience breakdowns. Treat the score as a signal, not a verdict. |
| TweetScout and similar X reputation tools | Shows who follows the account, which is more useful than how many. |
| Kaito and LunarCrush | Whether the account actually drives crypto conversation rather than just posting into it. |
| Arkham, Nansen, Etherscan, Solscan, Bubblemaps | Wallet history and token flows when a public address is known. |
| Wayback Machine | Finding promotions that have since been deleted. |
The 10-point checklist
Score every candidate out of ten. Anything below seven should not be booked, and anything that fails points 8, 9 or 10 should be cut regardless of the total. The KOL vetting scorecard is this checklist as a form you can fill in per account.
- The first fifty followers look like real accounts with posts and histories.
- Follower growth has no unexplained vertical spikes.
- Average engagement rate over the last ten posts sits inside the band for their tier.
- Replies outnumber a token count and include disagreement, not just emoji.
- The same real handles engage across several weeks.
- Audience geography matches the market you are selling to.
- Average views on the last ten videos or posts are within range of the quoted rate.
- Past promotions do not follow a pattern of peaking and collapsing within 48 hours.
- They disclose paid posts, and they do not delete them afterwards.
- They will show native analytics on a call and answer questions about results from their last three paid posts.
Once the list is clean, price it using crypto KOL rates. If you want to know what the outright scams look like rather than the merely weak accounts, read KOL scams and red flags.
Questions
What is a good engagement rate for a crypto KOL?
It depends on size. Under 10,000 followers, 2% to 6% is healthy. At 50,000 to 250,000 followers, 0.6% to 2% is healthy, and above 1 million, anything over 0.15% is normal. Rates far above the band usually mean bought engagement.
How do you calculate engagement rate?
Add likes, replies, reposts and quotes on a post, divide by the follower count, and multiply by 100. Average that across the last ten non-promotional posts rather than using the best one.
How can you tell if a crypto KOL has fake followers?
Read the first fifty followers for empty accounts created in the same month, check the follower growth chart for vertical spikes, and compare the reply count to the like count. Thousands of likes with almost no replies usually means bought engagement.
Should I check a KOL's wallet?
Yes if they have a known public address, but only as a disqualifier. Finding a wallet that received a project allocation and sold it straight into the promoted post is a reason to walk away. Finding nothing proves nothing, since most use several wallets.
What if a KOL refuses to show their analytics?
Cut them. Every honest account will screen share the native analytics tab on a call. A refusal, or a static screenshot instead of a screen recording, is the most reliable single red flag there is.
How long does vetting one KOL take?
About fifteen minutes per account once you have a process. On a longlist of 60 accounts that is roughly two days of work, and it usually removes half the list.
