Instagram Fake Follower Auditor

Audit public Instagram profiles for audience quality and likely fake-follower risk. The Actor reads public profile counts and recent-post engagement, then returns one dataset row per handle with an explainable risk score, confidence, risk tier, findings, and a recommended next step. It runs cookieless, with no Instagram login, and is built for humans, APIs, and AI agents on Apify.

Designed for agencies, brand teams, and CRM workflows that want a structured audience-quality score before a paid collab or influencer brief.

Best fit

• You have public Instagram handles or profile URLs and want a fake-follower risk score with reasons.
• You want confidence-gated tiers (LOW / MEDIUM / HIGH / VERY_HIGH / INCONCLUSIVE) instead of a raw follower count.
• After a niche shortlist, start with Instagram Niche Influencer Finder, then pass chosen handles here to compare audience quality.
• For bio, followers, and parsed contacts, continue with Instagram Profile Scraper or Instagram Contact Scraper. This Actor stays focused on public-signal authenticity scoring.

Workflow: from a shortlist to a collab decision

An agency starts with three public handles from a brief. It then runs this Actor with profiles: ["creator_a", "creator_b", "creator_c"] and maxRecentPosts: 12. Next it exports the overview view, compares weightedRiskScore, riskTier, and recommendation, and feeds HIGH-tier names to a human reviewer before spend. LOW-tier names can pass to outreach. The same flow works on a schedule for marketplace creator intake.

How scoring works

1. Fetch the public profile (followers, following, posts, private/verified flags).
2. Sample recent public posts (likes and comments).
3. Score five public-only signals, weighted:
   • Engagement-rate vs audience-size band (35%)
   • Like/comment balance (25%)
   • Follower-to-following pattern (20%)
   • Content depth vs claimed audience (10%)
   • Engagement volatility (10%)
4. Gate the tier with confidence. Thin evidence is labeled INCONCLUSIVE so the score stays honest.

Mega accounts are compared to mega-account engagement floors, so a large publisher is scored against its own band rather than a nano-creator rate.

Input

Field	Type	Default	Description
profiles	array (required)	["natgeo"]	Public handles, @handles, or profile URLs (up to 50 per run)
maxRecentPosts	integer	12	Recent posts in the engagement sample (3–24)
minConfidenceToClassify	number	0.55	Below this, the tier is INCONCLUSIVE
providerOrder	enum	scrapecreators-first	Public-data provider priority; fallback fires automatically
includePostLevelDiagnostics	boolean	false	Attach sampled post like/comment counts

Quick start input

{
  "profiles": ["natgeo"],
  "maxRecentPosts": 12,
  "minConfidenceToClassify": 0.55,
  "providerOrder": "scrapecreators-first",
  "includePostLevelDiagnostics": false
}

That input audits one well-known public profile so the quality run stays short and still returns a real scored row.

Output

One dataset item is one profile audit. Download JSON, CSV, Excel, or HTML from the dataset tab. Every run also writes OUTPUT and RUN_SUMMARY with the outcome, row counts, and charged events.

{
  "inputProfile": "natgeo",
  "normalizedHandle": "natgeo",
  "profileUrl": "https://www.instagram.com/natgeo/",
  "auditTimestamp": "2026-08-18T11:00:00.000Z",
  "outcome": "COMPLETE",
  "source": "scrapecreators",
  "weightedRiskScore": 18.4,
  "confidence": 0.9,
  "riskTier": "LOW",
  "engagementRatePct": 0.0746,
  "keyFindings": ["No major fake-follower anomalies in the available public signals."],
  "recommendation": "Low public fake-follower risk. Visible engagement is consistent with the reported audience size.",
  "billingEvent": "profile-audited"
}

Field	What it answers
normalizedHandle + profileUrl	Which public profile was audited
weightedRiskScore + riskTier	How the public signals score on fake-follower risk
confidence	How much public evidence supported the tier
engagementRatePct	Average likes plus comments vs follower count
keyFindings + recommendation	Why the score looks this way, and a suggested next step
outcome	COMPLETE, PARTIAL, VALID_EMPTY, INVALID_INPUT, UPSTREAM_FAILED, or CONFIG_ERROR

Pricing

This Actor uses Pay per event plus Apify platform usage. Open the live Pricing tab for current event details, and use Apify's run cost controls to keep the workflow aligned with your budget.

Event	Price	Charged when
Actor start	$0.00005	Once per run (scaled by memory)
profile-audited	$0.02	Each COMPLETE or PARTIAL audit row

Run	Audited rows	Event cost (approx.)
1 public profile	1	~$0.02
5-handle shortlist	5	~$0.10

VALID_EMPTY, INVALID_INPUT, and upstream-outage rows are saved when useful and stay outside profile-audited. Platform usage is billed on top at Apify's rates — the live Pricing tab is the current source of truth.

API example

curl -X POST "https://api.apify.com/v2/acts/khadinakbar~instagram-fake-follower-auditor/runs?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"profiles":["natgeo"],"maxRecentPosts":12}'

Then read the dataset and the OUTPUT record for outcome, row counts, and charged events.

AI agent usage

Audit public Instagram audience quality before a paid collab. Tool: khadinakbar/instagram-fake-follower-auditor Input: {"profiles":["natgeo"],"maxRecentPosts":12} Read the dataset row for weightedRiskScore, riskTier, confidence, recommendation, and outcome. Treat INCONCLUSIVE as a cue to gather more evidence. Cost is $0.02 per COMPLETE/PARTIAL row plus platform usage.

Best results

• Provide currently public handles; public posts are the engagement sample.
• Keep maxRecentPosts at 12 unless you need more evidence for a borderline account.
• Read keyFindings together with the risk tier before a spend decision.
• Pair a LOW or MEDIUM score with a live content check, then continue outreach.

Builder's note

I built this after live probes of native Instagram HTML and two public-data vendors. Direct instagram.com fetches returned empty shells, while ScrapeCreators /v1/instagram/profile plus /v2/instagram/user/posts (and SociaVault scrape equivalents) returned usable follower counts and post like/comment totals. I learned that mega-publisher engagement rates look “low” if you reuse nano-creator floors, so the scorer uses audience-size bands and confidence gates. I designed the billed unit as one public-signal audit row, because this contract scores profile and post metrics and leaves follower-identity sampling to other tools.

Responsible use

Use this Actor on public Instagram profiles you are authorized to research for legitimate due diligence, influencer vetting, or marketplace intake. It reads public profile and recent-post metrics only. Follow Instagram's terms of service, GDPR, CCPA, and applicable laws. The score is a probabilistic public-signal indicator for research, not a legal finding.

Changelog

• 0.1 — Initial private release. Provider-backed public audit, explainable scoring, PPE + usage, MCP-ready schemas.