Vet before you pay, not after
A follower count tells you almost nothing. AI can run the shortlist and structure the vetting sheet; the real follower, engagement and audience numbers have to come from the platform's own creator insights or a discovery tool (Modash, Heepsy) — never from a guess. Run every creator through the same short filter before money changes hands.
Fastest path: one prompt, end to end
🤖 AI prompt — paste into ChatGPT / Claude
You are my creator-vetting analyst. I'm considering these creators: [paste handles / profile URLs]. Product: [product]. Target buyer: [who buys it].
For EACH creator, open their public profile (and a discovery tool — Modash, modash.io, or Heepsy, heepsy.com — if you can) and pull REAL numbers:
1. Follower count and tier (nano 1K–10K / micro 10K–100K / macro 100K–1M / mega 1M+).
2. Engagement rate, and whether it sits inside the normal band for that tier (nano 5–9%+, micro 2–6%, macro ~2–2.5%, mega often <2%). Flag any account sitting well BELOW its tier — that's the red flag.
3. A quick fake-audience read: are the comments specific and conversational, or generic ("🔥 nice!") / bot-repetitive? Any sudden unnatural follower spike?
4. Audience fit: does their follower demographic (age, geo, language) match my target buyer?
5. If you can find a public post with a view count, only cite it as evidence if it has ≥10,000 views.
Output ONE markdown table: Handle | Tier | Followers | Engagement % | In-band? | Fake-audience flag | Audience-fit (Y/N) | Verdict (shortlist / pass).
Do not guess or estimate any number. Every figure must come from a profile you opened or a tool you can cite — put the source next to it. If you cannot get a real number, leave that cell blank and tell me which tool to open. If you can't browse, return the empty table with column headers and the tool names to use.
Or do it in 5 steps
- Source a first list fast. Platform hashtag/sound search, a discovery tool (Modash, Heepsy) or Shopify Collabs for demographic + fraud data, and — most underused — your own customers and DMs, plus competitor comment sections.
- Check engagement against the tier band, not in absolute terms. An account well below its tier's normal range is the warning sign. Read the comments: a real audience argues, asks "where do I buy this," tags friends; bots post repetitive "🔥😍" on every post.
- Look for a bought-follower spike. A free growth-chart tool (Social Blade) catches a sudden unnatural jump before you pay for a full audit. This matters — HypeAuditor found 49% of Instagram influencers worldwide had used fake followers at some point, a problem estimated to cost brands around $1.3 billion (Digiday).
- Match audience to buyer. Pull the age/gender/geo/language breakdown from creator insights or a vetting tool. A smaller creator whose audience is 90% your buyer beats a bigger one at 20%.
- Put it in writing. Deliverables, usage-rights duration, exclusivity, payment terms, and an attestation the creator hasn't bought followers or engagement — a clause more agencies added since the 2025 fraud scrutiny. Bake #ad/#sponsored into the brief as a non-negotiable line.
Done looks like: a shortlist sheet where every kept creator has a real in-band engagement rate and a documented audience-fit call.
Worked example: vetting sheet (fill with real numbers)
| Handle | Tier | Followers | Engagement % | In-band? | Fit | Verdict |
|---|
| @example_a | micro | (from tool) | (from tool) | ✅ 4.1% in 2–6% | Y | shortlist |
| @example_b | macro | (from tool) | (from tool) | ❌ 0.8% below ~2% | N | pass |
The tier bands are rough industry ranges, not a hard-sourced figure — read them as directional; the numbers in each row must come from the profile or tool you opened, not from this table.
Re-vet before every new campaign — audiences and engagement drift.