The Fraud Checks Brands Run Before Signing a Creator
What brand-side creator vetting actually looks like: the fraud data, manual checks, AI-assisted discovery, the FTC backstop, and the creator's side.
Every brand that has run an influencer program long enough has a story about the campaign that underperformed for a reason that had nothing to do with the creative: the account had the following, but not the audience. Vetting a creator before signing them is not paperwork bolted onto a partnership — it's the part of the deal that determines whether the number on their media kit means anything at all.
The fraud math, in one breakdown
Industry benchmark data puts a shape on the problem that's worth internalizing before anything else: fake or bot followers account for 56.5% of all reported influencer fraud and quality issues — more than five times the next category. Inauthentic or templated comments add another 10.6%, and outright purchased engagement adds 10.2% on top of that. Fake followers aren't one fraud vector among several roughly-equal ones; they're the dominant failure mode by a wide margin, which is exactly why the first vetting question should be about whether the audience exists, not about how enthusiastic it looks.
The legal backstop
This stopped being purely a brand-side risk-management question in August 2024, when the FTC finalized its rule banning the sale or purchase of fake indicators of social media influence — bot followers, purchased views, and similar manufactured signals — as a form of commercial misrepresentation, with civil penalties available for violations. The rule targets the transaction itself, not just the claims built on top of it: buying the followers is the violation, independent of what gets said about them afterward. For a brand, that reframes vetting from "protect the campaign's performance" to "protect the company from a compliance exposure it didn't create but sits directly next to" — a reason to document the check, not just perform it informally.
The manual checks
Automated tools help, but the checks that catch the most obvious fraud are still ones a person can run in twenty minutes:
Engagement authenticity. Comments that are generic and interchangeable across posts — the kind of line that could sit under any photo from any account — are a stronger signal than a raw engagement-rate number, because engagement rate is exactly the metric purchased engagement is bought to inflate. Comments that reference the actual content of the post are the harder signal to fake at scale.
Audience geography. A creator whose stated market and content language is one country but whose follower base skews heavily toward regions with no obvious audience fit is a classic tell — not disqualifying on its own, but a prompt to ask why before a brief and a fee are on the table.
Growth curve shape. Organic follower growth is bumpy and roughly continuous. A vertical jump on a specific date followed by a flat plateau is the shape purchased-follower batches leave behind — it's visible on most platforms' own public growth charts without needing a third-party tool at all.
Comment-to-follower ratio over time, not as a snapshot. A single viral post skews any ratio measured in isolation; a pattern across the last ten to twenty posts is a much harder thing to manufacture consistently.
Disclosure habits on existing sponsored posts. How cleanly a creator already flags paid partnerships is a reasonable proxy for how they'll handle disclosure on your campaign — an account that's sloppy or evasive about past sponsorships now is a preview of a compliance problem later, not a one-off oversight.
Who actually runs the check
In practice this vetting rarely has a dedicated owner — it falls to whoever is already accountable for the relationship, which on most teams is the brand manager, not a separate compliance function. That's worth naming explicitly, because the brand manager's usual job is keeping every asset that touches the brand on-model — locked reference sets, an approved look, consistent voice — and a creator partnership is an asset in exactly that sense: something the brand's name gets attached to whether the underlying audience is real or not. Folding fraud checks into the same review that already covers brand fit means the question gets asked before a contract exists, not after a campaign underperforms and someone goes looking for why.
Where AI-assisted discovery fits
None of these checks are new, but the volume of creators most programs now screen has outgrown doing them one profile at a time — which is why creator discovery is the leading AI use case in influencer marketing today, cited by 36.67% of marketers as where AI actually earns its place in the workflow, ahead of content generation or campaign reporting. The job AI-assisted discovery does well is exactly the pattern-matching above, run at scale: flagging growth-curve anomalies, scoring comment authenticity across a full post history, and surfacing audience-geography mismatches automatically instead of by hand — turning a twenty-minute manual check into a first-pass filter across hundreds of candidates, with the human check still applied to whoever clears it. AI adoption in the discipline overall is broad enough that going without any of it is now the minority position: only 10.56% of marketers report using no AI at all anywhere in their influencer workflow.
There's a second, less obvious way generated video changes this calculus: some of what a creator relationship is actually used for — testing whether an angle or a hook works before committing budget to it — doesn't require a creator at all in the testing phase. Versely's guide for influencer marketing agencies is built around exactly that reordering: render a rough version of the brief and test hook variants in paid before a creator is ever booked, so the fraud-vetting question only has to be asked about the person you're actually about to pay, not about every candidate you're still comparing.
The flip side: proving authenticity in a pitch
Creators reading fraud coverage from the buyer's side should take the same checklist as pitch prep, not as a threat. The pitches that clear vetting fastest lean on things that are hard to fake rather than things that are easy to state:
- Native platform analytics screenshots, not a self-reported media kit number. A follower count is a claim; a platform's own insights screen showing consistent engagement over months is evidence.
- A trend, not a spike. One viral post is exciting and proves nothing about the account underneath it — a brand vetting seriously is looking at the ten posts before and after the spike, not the spike itself.
- Offering a small paid test before asking for a full retainer. Versely's guide to UGC ads for brands frames the creator-brand relationship as a supply arrangement, not a vibes-based one — a small first deliverable that performs is a better opener than any number on a rate card, because it's the one thing that can't be purchased in bulk from a bot farm.
Why this matters more than it looks like paperwork
The entire value proposition of a UGC ad is that it reads as a real recommendation from a real relationship between a creator and their audience — that's the whole reason the format outperforms a polished commercial in the placements where it runs. Fake followers and purchased engagement don't just waste a media budget; they counterfeit the exact thing the format is being bought for. Vetting a creator isn't due diligence layered on top of the campaign — it's confirming the product you're actually buying exists in the first place. Brand teams building this check into a standing process, not a one-off before a big spend, are the ones who stop relearning this lesson every quarter.