TikTok Influencer Audit — Fake Followers, Engagement & Reach

Vet a TikTok creator before you pay them. Give handles and get the full agency due-diligence brief for each: engagement benchmarked to the creator's follower tier, measured both by followers and by views; fake-follower and bought-engagement signals (followers who don't watch, lifetime likes vs. followers, like/view/comment ratios that don't add up); reach health and momentum (is the audience watching now?); brand-safety flags; a CPM-based rate card; and authenticity, reach-health and influence scores with a verdict and a one-line summary. A TikTok scraper runs under the hood, so you only pass handles. Agent-callable.

Built for influencer-marketing agencies, brand partnership teams, TikTok Shop affiliate managers and creator marketplaces that lose budget to inflated follower counts.

Tiers (deep layers run only when you ask)

• Profile audit (default, ~20 s): tier-benchmarked engagement (by followers and by views), bought-follower tells, reach health + momentum, cadence, sponsored share, brand safety, rate card, three scores, verdict.
• Comment forensics (deepComments: true): reads comments on recent videos and grades how authentic the engagement is: % generic/emoji-only, spam/solicitation, duplicate text, commenters still on TikTok's auto-generated user123… handle, and comment-language mismatch. The result is a commentAuthenticityScore blended into the verdict. A second scraper runs, so it takes longer; best for scheduled, webhook or batch runs.

These are authenticity signals computed from public creator data for due diligence. They are not an accusation that a creator is fraudulent. Benchmarks and the rate card are indicative. Verify before contracting.

What makes it deep (vs single-number tools)

Dimension	What it computes
Engagement vs tier	ER = median (likes+comments+shares) / followers, scored against the norm for the creator's tier (nano → global). Also ER by views, the TikTok-native quality measure.
Do the followers watch?	Median views ÷ followers vs the tier norm. TikTok pushes videos to non-followers, so an account whose followers never watch is the classic bought-follower pattern.
Lifetime likes per follower	Organic TikTok accounts earn many lifetime likes per follower. A follower count larger than every like the account ever received points to bought followers.
Ratio forensics	Like-to-view (bought likes / botted views), comments per 1k likes (lopsided = bought likes; extreme = giveaways/pods), view-count uniformity.
Reach health + momentum	Recent likes per video vs the creator's own lifetime average (rising / steady / cooling / declining), plus posting activity. Don't pay for 2021's numbers.
Honest sampling	Pinned showcase videos (often 50–250× a normal video) and videos under 48 h old are excluded from the rates. Medians, not averages, so one viral hit can't inflate the read. Hit-dependence is reported.
Credibility gating	Verified/established accounts aren't false-flagged as fake for naturally low engagement. Their dormant followers show up as low reach health instead, which is accurate and fair.
Commercials	CPM rate card on median views × engagement quality (TikTok deals price on reach, not follower count), sponsored-video share, TikTok Shop activity.
Brand safety & fit	Risky-content flags (adult / gambling / crypto-promo / engagement-bait / profanity) and an optional brand-keyword fit.
Verdict	authenticityScore + reachHealthScore + influenceScore + signal + summary line + findings.

Verdicts: authentic-healthy-engagement · authentic-modest-engagement · inactive-audience-low-reach · inflated-followers-suspected · bought-engagement-suspected · engagement-authenticity-concerns · authenticity-concerns--review · brand-safety-review · private-account · not-found.

Why not just ask an AI agent?

An agent can open a profile page. It can't pull a creator's recent videos with their real view/like/comment/share/save counts, strip the pinned showcase hits that make every profile look bigger, benchmark against the right tier norm, run the bought-follower and bought-engagement forensics, and roll it into consistent scores for a whole roster in one call. That encoded influencer-marketing expertise at scale is the product.

Input

Field	Type	Required	Description
usernames	array of strings	✅*	Handles or profile URLs (zachking, @khaby.lame, https://www.tiktok.com/@duolingo).
brandKeywords	array of strings		Topics for a brand-fit read.
datasetId	dataset	✅*	…or read handles from another dataset (e.g. a TikTok search run).
videosSampled	integer		Recent videos analysed per creator (default 10).
deepComments	boolean		Comment forensics add-on (default off).
commentVideosSampled	integer		Videos sampled for comments (default 3).
commentsPerVideo	integer		Comments read per video (default 30).
maxCreators	integer		Cap per run (default 20).
webhookUrl	string		POST the audits here as JSON.

* provide usernames or datasetId.

Example input

{ "usernames": ["zachking", "khaby.lame", "duolingo"], "brandKeywords": ["comedy", "magic"] }

Output

One audit per creator. Example (real run, trimmed):

{
  "username": "zachking",
  "verified": true,
  "tier": "global",
  "followersCount": 86800000,
  "medianViews": 12450000,
  "viewToFollowerRatio": 0.143,
  "engagementRatePct": 1.472,
  "engagementVsBenchmark": 1.84,
  "engagementRateByViewsPct": 10.84,
  "likeToViewPct": 10.51,
  "commentsPer1kLikes": 5.6,
  "saveRatePct": 0.494,
  "lifetimeLikesPerFollower": 14.98,
  "momentum": {"label": "steady", "recentVsLifetimeLikesRatio": 0.65},
  "cadence": {"avgDaysBetweenVideos": 10.0, "daysSinceLastVideo": 17.8},
  "authenticityScore": 95, "reachHealthScore": 90, "influenceScore": 89,
  "estimatedRateCardUsd": {"low": 186750, "mid": 373500, "high": 653625, "currency": "USD"},
  "sample": {"videosAnalyzed": 8, "pinnedExcluded": 2, "freshExcluded": 0},
  "signal": "authentic-healthy-engagement",
  "summary": "@zachking - global tier (86,800,000 followers) - median 12,450,000 views (14% of followers) - ER 1.47% by followers / 10.8% by views - reach steady - authenticity 95, reach health 90, influence 89 -> authentic-healthy-engagement",
  "findings": [
    {"severity": "low", "code": "pinned-showcase", "detail": "2 pinned showcase video(s) at ~99x the median views were excluded - the profile page overstates typical reach."}
  ]
}

A verified account whose videos now reach under 1% of its followers returns inactive-audience-low-reach (real followers, but you'd be paying for an audience that no longer watches). An unverified account with followers far above its views and lifetime likes returns inflated-followers-suspected. Risky content returns brand-safety-review.

Use cases

• Pre-deal vetting: confirm a creator's audience and engagement are real, and currently watching, before you sign.
• Roster screening: audit a whole shortlist or a discovery dataset in one run and rank by authenticity and reach health.
• Rate negotiation: an independent, reach-based fair-price estimate.
• TikTok Shop affiliate selection: spot creators whose reach is current, not legacy.
• Brand safety: flag risky content and ad-fatigue before association.
• AI agents: a structured creator verdict (plus a one-line summary) to gate outreach.

For AI agents / MCP

https://mcp.apify.com/?tools=actors,docs,centorprat/tiktok-influencer-audit

Loads as centorprat--tiktok-influencer-audit. Claude Code:

claude mcp add apify -- npx -y @apify/actors-mcp-server --actors centorprat/tiktok-influencer-audit

FAQ

Where does the data come from? Public TikTok profiles and recent videos, via TikTok scrapers on Apify that run under the hood. You don't manage any key or login.

Why are pinned videos excluded? Creators pin their biggest hits. In our tests pinned videos ran 50–250× a creator's normal views. Leaving them in inflates every average, which is how a profile page misleads.

What's the difference between authenticity and reach health? Authenticity asks whether the followers and engagement are real. Reach health asks whether the followers watch now. A creator who went viral years ago can be 100% authentic and still a poor buy today.

Does it accuse creators of fraud? No. It surfaces signals for your review, and verified/established accounts are gated so they aren't flagged as fake for naturally low engagement.

How is engagement benchmarked? Against typical TikTok ranges for the creator's follower tier, both per follower and per view, using medians across recent videos.

Private accounts? Reported as private-account with public metrics only.

Does it look at audience nationality? No. The comment add-on only checks whether comment language matches the creator's own content (bot farms typically post in unrelated languages). Commenters are never named or profiled.

How fast is it? About 20 seconds for a profile audit. With comment forensics, about a minute.

Cost? Pay-per-event: one event per creator audited, plus one per creator when comment forensics runs. The underlying scraper usage is billed to your run: a few cents per creator for the audit. With deepComments, a TikTok comment scraper also runs, which adds up to about $0.11 per creator on Apify's Free plan (3 videos x 30 comments; much less on paid plans or with fewer comments).

Data & attribution

Public TikTok data via TikTok scrapers on Apify. These are signals for influencer due diligence, not accusations or guarantees. Rate-card and benchmark figures are indicative.

Related actors

• Instagram Influencer Audit
• YouTube Influencer Audit
• Google Maps Leads & Website Audit
• Shopify Store Leads & App Audit