Three crypto influencer case studies run in different markets and formats: a tiered TikTok package, a YouTube placement, and a Chinese-language campaign that ran on WeChat rather than any Western platform. The last one is the most instructive, because it is the case where none of the usual channels applied.
Mirrorworld: a tiered TikTok package
Four integrated placements structured by audience size rather than by name recognition. One creator at 100,000 to 200,000 followers, one at 50,000 to 100,000, and two between 10,000 and 40,000. The campaign ran across five to ten days including pitching, agreement and production of each video.
Tiering exists because reach and conversion do not move together. The largest account produces the impression count that goes in the report, and the smaller ones usually produce the comments worth reading. Buying only the top tier is how a campaign gets a good number and no conversation.
BitYard: a YouTube placement
A single video placement on an exchange campaign. YouTube is the slowest of the creator formats to report and the longest to keep working, because the video continues accumulating views for months after payment. Judging it against a TikTok campaign at the one-week mark will always flatter the TikTok.
Pirates World: a Chinese campaign on WeChat
A GameFi and NFT project needed investors in China and a Chinese-speaking community, which rules out every channel a Western campaign would use. A creator produced a video, and that video plus a QR code to the project site was distributed across more than 7,000 WeChat communities, supported by articles on 12 forum media outlets.
The QR code is the detail that matters. WeChat does not behave like an open web platform, and a campaign designed around clickable links simply does not function inside it. Regional campaigns fail more often on mechanics like this than on translation quality.
What these crypto influencer case studies say about creator selection
Format determines the reporting window, and market determines the mechanics. A tiered TikTok package reports in days, a YouTube placement reports over months, and a WeChat campaign reports in a way that Western analytics cannot see at all.
The vetting standard is the same across all three: engagement relative to audience size, and whether that engagement is genuine. Our active X roster averages 1.74 percent across its 85 accounts, and networks quoting several times that across hundreds of accounts are measuring something else. Our guide to influencer channels compared covers what each format actually buys.
For the current service, see crypto influencer marketing.
Planning a creator campaign? Send the brief on Telegram at https://t.me/flexe_io_agency and we will come back with a vetted shortlist within 24 hours.
Other case studies
Campaigns grouped by discipline: PR, paid advertising, community, forum and airdrop, content production.
How creators were selected in these campaigns
None of the three campaigns above was built by sorting candidates on follower count. In each case the shortlist came from the same checks, and in each case the accounts that survived them were smaller than the ones originally proposed.
- Median, not best. The middle value across recent posts, because one viral post distorts an average and the media kit is usually built around exactly that post.
- Paid against organic. Their previous sponsored posts compared with their normal content. A steep drop means the audience recognises promotion and will treat yours the same way.
- Reply quality. Specific questions from accounts with history, rather than short generic praise from accounts created in the same week.
- Promotion history. What they posted about over the preceding months, and what happened to those projects afterwards.
The full procedure, including what each check looks like in practice, is in how to vet a crypto KOL in 20 minutes.
What the briefs had in common
Each creator received facts and prohibitions, not a script. Product function, dates, chains, audit status, contract address and the figures that could be quoted, plus an explicit list of what could not be said: no price predictions, no return claims, no unconfirmed listings, no investment framing.
The phrasing was left to the creator. A post written in the project’s marketing register inside a feed the audience follows for a different voice reads as an advertisement whether or not it is labelled as one. The brief format is set out in how to brief a crypto KOL.
What was measured, and when
Every creator had a distinct tracked link built before the brief went out, which is what makes any of these results attributable at all. Post metrics were captured at twenty four hours and again at seven days, because comparing creators requires reading numbers at the same age.
The measurement that changed decisions was the thirty-day one. Retention differs sharply between creators with similar reach, and in more than one of these campaigns the ranking at thirty days reversed the ranking at twenty four hours. What to record and in what order is in how to measure whether a KOL campaign worked.
The accounts behind campaigns like these are compared in our top crypto influencers list, with verified followers and median views per post.
Reading crypto influencer case studies without copying
Every published case carries the conditions it ran in, and those conditions are usually missing from the write-up. The market that month, the token, the competing launches in the same week, and the state of the project’s own audience all shaped the numbers that appear, and none of them transfer. What does transfer is the decision sequence, which is why the useful part of a case is the part describing choices rather than outcomes.
Read a case for what it declines to claim. A write-up reporting impressions and nothing after them has told you the campaign was measured at the top of the funnel only, which may be all it was built to do. A write-up reporting a figure without the window it was measured over has told you less than it appears to. Neither is dishonest, and both change what you should expect from a similar booking.
Building the shortlist for a new campaign
Start from the conversation rather than from a list. Find the accounts being quoted and argued with on the specific subject the campaign is about, then work outward through whoever replies to them with something worth reading. The names arriving this way are smaller than the ones an agency list would propose, which matches what happened in the campaigns above, where the accounts that survived the checks were smaller than those first suggested.
Run the checks in an order that fails candidates cheaply. Median views first, because it removes the most names for the least effort, then paid against organic, then the replies, then promotion history, which takes the longest and is worth doing only on the accounts still standing. Reversing that order means reading months of posts for accounts a single median would have excluded in a minute.
Decide the tier mix before pricing anything. The campaigns above were structured by audience size rather than by name recognition, with one larger account and several smaller ones, because reach and conversion do not move together. The larger account produces the figure that goes in the report and the smaller ones usually produce the replies worth reading, so buying only the top tier buys a number and no conversation.
Keep the rejected names with the reason attached. A shortlist is built once and used for years, and a record of why an account was excluded saves the next campaign from proposing it again. It also catches the case where an account was rejected for something that has since changed, such as a growth spike that has settled or a promotion history that has aged out.
Running a regional campaign end to end
A campaign outside the platforms you know starts with mechanics rather than with creators. Establish first how content moves in that market, what a viewer can tap, whether links function at all, and how an interested reader reaches a destination. The WeChat campaign above worked because distribution used a QR code rather than a link, which is a mechanical decision made before any creator was approached.
Plan for the fact that you cannot read the result directly. Figures from a closed platform arrive through the partner, in the partner’s format, and often without the detail you would take for granted elsewhere. Agree the reporting format before the campaign starts, ask for screenshots at fixed points, and set expectations internally that the reading will be coarser than a Western campaign of the same size.
Translation is the smallest part of the work and the part most often treated as the whole. A translated script delivered in the register of the original still reads as foreign material, and the audiences that matter notice immediately. Give the local creator the facts and the prohibitions and let the phrasing be theirs, the same way the briefs in these campaigns did.
Production timeline and approval steps
Work backwards from the publication date and leave room for one revision. A creator needs time to produce and publish, and the approval step in the middle is where campaigns lose their dates, usually because the review sat with someone who had not seen the brief. Name the reviewer before the brief goes out and give them a single window rather than an open invitation to comment.
Limit the review to facts and prohibitions. Correcting a contract address, a date, or a claim that cannot be made is the reviewer’s job; rewriting the creator’s phrasing is how a post stops sounding like the feed it appears in. A review returning the draft covered in style changes also costs the relationship, and the next quote from that account will reflect it.
Agree what happens when the draft arrives late. A creator who misses the review window with the publication date fixed leaves two options, publishing unreviewed or moving the date, and the choice is easier if it was made in advance. For a launch tied to an external event, the honest answer is often to drop that creator from the wave rather than to compress the review.
Proof of publication and payment terms
Collect proof at publication rather than at the end of the campaign. A link, a screenshot with a visible timestamp, and the early metrics from the creator’s own view take a minute each on the day and are difficult to reconstruct weeks later, especially where a post has since been edited or removed. Store them with the campaign record rather than in a message thread.
Structure the payment around the deliverable, not around the result. A creator controls what they publish and cannot control how it performs, so tying a fee to performance either prices in the risk or attracts accounts confident about numbers for reasons you would not like. Where an incentive is wanted, put it in the next booking rather than in the current fee.
What a campaign report should contain
A report another person can act on records the decisions alongside the numbers. Which accounts were considered and rejected, what each was paid, what the brief allowed and prohibited, when each post ran, and the readings taken at each fixed interval. Without the rejected names and the brief, a strong result is unrepeatable, because nobody can tell which part of it produced the outcome.
Separate the readings by age rather than merging them into a campaign total. A total mixes a format reporting in days with one reporting over months, and it hides the reversal that matters, where the ranking after thirty days differs from the ranking on the first day. Reporting both readings side by side is what makes the second campaign better than the first.
Keep the assets and the raw exports, not only the summary. Copies of each published post, the files the creator used, the tracked link definitions, and the exports behind every figure are what let a result be defended months later when somebody asks how it was calculated. Campaign records consisting of a slide with a few numbers on it are why the same questions get re-answered every quarter.
When a campaign goes sideways
The common failures are procedural. A post published without the tracked link, a contract address typed wrong, a video removed by the platform over a claim in the script, or a creator who publishes early into a market that has moved. Each of these is recoverable inside a day if it is noticed, and expensive if the check happens at the end of the week.
Assign someone to watch the first hour of each publication. Reading the replies early catches a factual error while a correction still looks like diligence rather than damage control, and it catches the case where the audience has understood the offer differently from how it was written. The same hour tells you whether the next creator in the wave needs a changed brief.
Where a campaign underperforms across every account, suspect the message before the accounts. Several screened creators producing weak results at the same time points at what they were asked to say, the destination it pointed to, or the timing, rather than at a simultaneous failure of judgement by people selected on separate evidence. Rebooking different names with the same brief repeats the result.
What transfers from these crypto influencer case studies to a different project
The method transfers; the numbers do not. A campaign result depends on the token, the market conditions in that month, the landing destination and the audience match, and quoting another project’s engagement figure as an expectation is how briefs acquire targets nobody can hit.
What is portable is the sequence: vet on medians and replies, brief on facts and prohibitions, give each creator a unique link, and judge at thirty days rather than at twenty four hours.
Screened accounts, negotiated terms and campaign reporting are described on our crypto influencer marketing page.