Crypto KOL campaign benchmarks: users, volume and CPA from campaigns we reported

Why most crypto KOL benchmarks are not benchmarks

Search for crypto KOL benchmarks and you get three things: rate cards by follower tier, engagement-rate rules of thumb, and vendor analyses of their own data. All three are useful and none is a campaign result. A rate card tells you what a post costs, not what it produces. An engagement heuristic tells you whether an audience is awake, not whether it buys. A vendor analysis tells you about the vendor's dataset.

Cookie3's June 2026 analysis of 175,370 active crypto KOLs is a good example of the third kind. It reports that more than half of those accounts produce zero quality-weighted engagement per post, that accounts over 500,000 followers have a far lower median signal per post than much smaller investor and builder accounts, and that under a fifth of scored accounts are classed as clean of farming behaviour. Those are the vendor's own proxy scores on an unstated window. They are directionally consistent with what we see. They are not conversion rates.

Campaign results are rarer because most agencies do not publish them. The figures below are ours, from our cases page, with the ratios worked out and the caveats attached.

Impressions to users: two exchange campaigns

For Binance in 2024, KOLs on X, YouTube and TikTok produced 732,000 impressions, 65,223 engagements and 4,245 new users. For MEXC in 2024, creators on X and YouTube produced 540,426 impressions, 75,450 engagements and 8,522 new users.

Three times the users per impression on the second campaign, with a smaller budget of impressions. The difference was not the platform mix, which overlapped, and not the offer type, since both used contests or competitions. It was the roster: MEXC's creators were booked for education first, promotion second, to an audience of active traders. The signals that predict which KOLs convert are the explanation for the gap.

Use the range, not either number. A crypto exchange KOL wave producing between roughly 5 and 15 users per 1,000 impressions is inside what we have seen. Below that, check the roster and the offer. Above it, check the traffic.

Volume per user: the number that warns you about averages

Binance reported more than $510M of trading volume from 4,245 new users, which is roughly $120,000 per user. MEXC reported more than $300M from 8,522 users, roughly $35,000 per user. Gate.io, measured on monthly volume rather than users, averaged $153M a month over three months against a $100M target.

The Binance-to-MEXC gap in volume per user is three and a half times. That is not a benchmark. It is a warning that volume is produced by a minority of accounts and that the average hides them. A campaign that brings 500 users including ten whales looks better on volume per user than one that brings 5,000 ordinary traders, and worse on almost every other measure.

If volume is the target, report the median per account and the top-ten concentration beside the total, as our guide to KOLs that bring trading volume sets out. If a vendor quotes you volume per user without those, they are quoting whales.

Targets beaten: what a defined job produces

Several of our campaigns were set a number before launch, which makes them the cleanest benchmarks we have, because the denominator was agreed in advance.

The pattern across them is not that targets are easy. It is that a campaign built around one defined action, with a roster cast for it, tends to beat a target set without creators in mind. The targets were set by clients from their own prior channels. Treat the share-of-target column as evidence that defined jobs work, not as a multiplier to apply to your own target.

Acquisition benchmarks for apps, wallets and testnets

Outside exchanges, the reported figures are counts rather than ratios, because the clients' cost and revenue lines are theirs. They still set a scale for what a creator campaign for a product launch produced.

• Swissmoney, crypto banking: more than 32,000 users onboarded, with cost per acquisition cut four times against the previous channel mix. This is the one figure on our cases page that is a CPA benchmark, and it is relative, not absolute.
• OKX, exchange network launch: 7,609 testnet participants from creator explainers, podcasts and X Spaces.
• io.finnet, MPC wallet infrastructure: more than 5,000 developer sign-ups in three weeks from technical KOLs.
• Illuvium, Web3 game: 6,550 new players and more than 24,000 followers from gaming creators, streams and events.
• STEPN: 5,344 new users and 949 NFTs sold from creators on X and YouTube.
• Polkadot Hydration, DeFi: 980 wallets and 5,420 followers from crypto KOLs.

Track360's published range for a mid-tier KOL is 100 to 500 wallet connections per campaign, with its community funnels converting 15-30% from join to wallet connection and 5-15% from connection to first deposit or trade. Read against that, 980 wallets from a KOL campaign on a niche DeFi protocol is a roster of several mid-tier accounts doing their job, and 7,609 testnet participants is a large wave with a strong explainer brief. Neither is extraordinary, which is the point of a benchmark.

For converting counts like these into a return, our crypto influencer ROI guide covers the model; for the cost side by tier, the KOL cost guide.

The denominator check, applied to every number above

A benchmark is a numerator over a denominator, and most arguments about benchmarks are arguments about the denominator nobody stated. Before accepting any figure, including ours, ask five questions.

1. What counted as a user? A registration, a KYC-passed account, a funded account, a first trade. Our exchange figures report new users as the client defined them; they are not all funded accounts, and we say so.
2. What was the window? Volume over a campaign month is not volume over a quarter. Gate.io is a three-month average; Binance and MEXC are campaign totals.
3. Attributed or total? Our reported users and volume are campaign-attributed through the client's tracking, which excludes people who saw a post and registered another way.
4. Who set the target? A target set by the client from prior channels is a fair denominator. A target set by the agency after the result is not.
5. Is it a count or a ratio? Counts scale with budget. The ratios above, users per 1,000 impressions and result as share of target, are the ones to compare across campaigns of different sizes.

Apply the same questions to the Octo Media casino streamer figures in our casino streamer conversion guide, to Track360's medians, and to Cookie3's signal scores. Every one of them answers a specific question on a specific dataset.

How to benchmark your own campaign against these

Take your last creator campaign and compute three ratios: users per 1,000 impressions, result as a share of the target set before launch, and the median per-account value beside the total if the action was volume or revenue. If you did not set a target before launch, the second ratio is unavailable, which is itself the finding.

Then compare like with like. An exchange campaign against the Binance and MEXC rows. A testnet against OKX and Saakuru. A DeFi protocol against Hydration. A product with a CPA from other channels against Swissmoney's relative cut. A campaign that lands inside the range is normal. One that lands far outside it in either direction has a story in the roster, the offer or the tracking, and that story is worth more than the ratio.

Our KOL marketing guide walks through the six campaign decisions that produce these numbers in the first place.

What to bring to the first call

Your last campaign's spend, impressions, engagements and attributed users or actions, with the definition of each. The target, if one was set. The product and the action you want next. From that we show you where you sit against the comparable rows above and scope the next wave in the crypto KOL marketing service, with a target agreed before launch so there is a denominator next time.