Instagram Creator Analyzer — Engagement, Niche & Sponsorship Data API for AI Agents

Instagram creator and influencer analytics without login, cookies, or an official API key. Give it usernames or profile URLs; it returns typed JSON rows — schema below — with profile stats (followers, following, posts), engagement rate, content mix, posting cadence, detected niche, and a per-post sponsorship value estimate. Extraction survives Instagram changes by falling back across 7 independent sources (Google, Social Blade, Instagram direct, Ninjalitics, Bing, DuckDuckGo, Wayback Machine) and merging the results with outlier rejection — most single-source Instagram scrapers break the moment one site changes its markup.

Why AI agents use this Instagram creator analyzer

• Deterministic typed output — every row follows the dataset schema (profile / analysis / meta objects); no HTML, no free-text parsing needed.
• Honest degradation — when post data cannot be fetched, posting.frequency is "unknown" and engagement-based sponsorship adjustments are skipped, never faked. meta.dataCompleteness (0–100) tells you exactly how much was recovered.
• Clear error semantics — failures are per-profile rows with analysis: null and a Fatal error: ... warning; the per-method attempt log (meta.methodResults) shows what was tried and why it failed.
• No auth needed — no Instagram account, session cookies, or API keys. Multi-source fallback keeps working when any single source blocks.
• Rate-limit handling built in — randomized delays between profiles and sources, per-method and per-profile time budgets, rotating browser fingerprints.

Input schema

Field	Type	Required	Default	Description
usernames	array of strings	yes	–	Instagram usernames or profile URLs to analyze (e.g. "natgeo" or "https://instagram.com/natgeo")
includeRecentPosts	boolean	no	true	Fetch and analyze recent posts for engagement metrics
maxPostsPerProfile	integer	no	12	Maximum recent posts to analyze (1–12)
proxyConfiguration	object	no	Apify proxy, RESIDENTIAL group	Proxy settings. Residential proxy is only spent on the Instagram-direct method (Instagram blocks datacenter IPs); all search-engine and aggregator fallbacks run over the direct connection. Residential proxy traffic is billed to your Apify account.

Minimal input:

{ "usernames": ["natgeo"] }

Output schema

One dataset row per input username. Row shape: { profile, analysis, meta }.

profile — raw merged profile data

Field	Type	Nullable	Description
profile.username	string	no	Handle, lowercase, no @
profile.id	string	yes	Instagram internal user ID (instagram_direct only)
profile.fullName	string	yes	Display name
profile.biography	string	yes	Bio text (may be a search-snippet approximation from fallback sources)
profile.externalUrl	string	yes	Link-in-bio URL
profile.externalUrls	array	yes	{ title, url } objects
profile.followersCount	number	yes	Followers (consensus across sources, outlier-rejected)
profile.followsCount	number	yes	Accounts followed
profile.postsCount	number	yes	Total posts (abbreviated counts like 30K are expanded to 30000)
profile.isBusinessAccount	boolean	yes	Business account flag
profile.businessCategoryName	string	yes	Business category
profile.isPrivate	boolean	yes	Private account flag
profile.isVerified	boolean	yes	Verified badge
profile.profilePicUrl / profilePicUrlHD	string	yes	Avatar URLs
profile.relatedProfiles	array	yes	{ username, fullName, isVerified } (instagram_direct only)
profile.latestPosts	array	yes	Up to 12 posts: { id, shortCode, type, caption, hashtags, mentions, likesCount, commentsCount, videoViewCount, videoDuration, timestamp, displayUrl }. Rich fields require instagram_direct; search fallbacks may yield shortcodes only.

analysis — derived analytics (null if the profile failed entirely)

Field	Type	Nullable	Description
analysis.audienceTier	string	via null analysis	nano | micro | mid | macro | mega | celebrity
analysis.engagement.rate	number	–	(avg likes + avg comments) / followers × 100. 0 when no post data exists — absence of data, not low engagement
analysis.engagement.avgLikesPerPost	number	–	Average likes per analyzed post
analysis.engagement.avgCommentsPerPost	number	–	Average comments per analyzed post
analysis.engagement.avgViewsPerReel	number	yes (omitted)	Average views across video/reel posts
analysis.engagement.likeToCommentRatio	number	–	Avg likes / avg comments
analysis.contentMix	object	–	{ images, videos, reels, carousels } percentages
analysis.posting.frequency	string	–	daily | every_few_days | weekly | biweekly | monthly | inactive | unknown (unknown = no post data available)
analysis.posting.avgPostsPerWeek	number	–	Average posts per week
analysis.posting.lastPostDate	string	yes (omitted)	ISO 8601 timestamp of latest analyzed post
analysis.posting.daysSinceLastPost	number	yes (omitted)	Days since latest analyzed post
analysis.niche	array	–	Up to 3 detected niches (fitness, beauty, tech, finance, ...)
analysis.shortsFocused	boolean	–	True if >50% of analyzed posts are reels
analysis.sponsorshipEstimate	object	–	{ low, medium, high, currency: "USD", adjustments }. Tier-based per-post rate with every applied adjustment listed as a string. When no post data exists, engagement adjustments are skipped and noted in adjustments.

meta — extraction metadata (always present)

Field	Type	Description
meta.scrapedAt	string	ISO 8601 row timestamp
meta.methodsUsed	array	Methods that succeeded: google, socialblade, instagram_direct, ninjalitics, bing, duckduckgo, wayback
meta.methodResults	array	Per-method attempt log: { method, success, data, error, latencyMs }
meta.dataCompleteness	number	0–100 field-coverage score; 0 = profile failed
meta.warnings	array	Per-method failure messages and fatal errors

Example row (successful profile)

{
  "profile": {
    "username": "natgeo",
    "fullName": "National Geographic",
    "biography": "See Instagram photos and videos",
    "profilePicUrl": "https://scontent.cdninstagram.com/v/...",
    "followersCount": 268796119,
    "followsCount": 194,
    "postsCount": 30400,
    "isVerified": true
  },
  "analysis": {
    "audienceTier": "celebrity",
    "engagement": { "rate": 0, "avgLikesPerPost": 0, "avgCommentsPerPost": 0, "likeToCommentRatio": 0 },
    "contentMix": { "images": 0, "videos": 0, "reels": 0, "carousels": 0 },
    "posting": { "frequency": "unknown", "avgPostsPerWeek": 0 },
    "niche": ["photography"],
    "shortsFocused": false,
    "sponsorshipEstimate": {
      "low": 25000,
      "medium": 75000,
      "high": 150000,
      "currency": "USD",
      "adjustments": ["no post data: engagement adjustments skipped"]
    }
  },
  "meta": {
    "scrapedAt": "2026-08-15T04:15:33.564Z",
    "methodsUsed": ["google", "socialblade", "instagram_direct"],
    "methodResults": [
      { "method": "google", "success": true, "data": { "username": "natgeo" }, "latencyMs": 1672 },
      { "method": "socialblade", "success": true, "data": { "username": "natgeo", "followersCount": 268796119, "postsCount": 30400, "followsCount": 194 }, "latencyMs": 190 },
      { "method": "instagram_direct", "success": true, "data": { "username": "natgeo", "fullName": "National Geographic", "followersCount": 268000000, "followsCount": 195, "postsCount": 30000, "isVerified": true }, "latencyMs": 15705 }
    ],
    "dataCompleteness": 84,
    "warnings": []
  }
}

Example row (failed profile)

{
  "profile": { "username": "some_unreachable_user" },
  "analysis": null,
  "meta": {
    "scrapedAt": "2026-08-15T04:20:00.000Z",
    "methodsUsed": [],
    "methodResults": [],
    "dataCompleteness": 0,
    "warnings": ["Fatal error: All extraction methods failed"]
  }
}

Error semantics

• Bad input (missing/empty usernames): the run logs an Invalid input error and produces zero dataset items. Treat a run with zero items as failed.
• Invalid username format: skipped with a log warning; no row is produced for it.
• Per-profile failure: the run continues to the next username; the failed profile still gets a row with analysis: null, meta.dataCompleteness: 0, and the error in meta.warnings. Retry logic: retry only usernames whose row has dataCompleteness === 0.
• Partial data: sources that failed are listed in meta.warnings / meta.methodResults; the row is still produced from whatever succeeded. Check meta.dataCompleteness before trusting derived analytics.
• Zero results: a run where every profile failed still completes; detect via item count or dataCompleteness values.

Use from AI agents (MCP)

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com/?tools=apricot_blackberry/instagram-creator-analyzer",
      "headers": { "Authorization": "Bearer <YOUR_APIFY_TOKEN>" }
    }
  }
}

Works in Claude, Cursor, ChatGPT deep research connectors, and any MCP client; the input schema above is the tool's parameter schema.

Use from code

curl:

curl -X POST "https://api.apify.com/v2/acts/apricot_blackberry~instagram-creator-analyzer/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"usernames": ["natgeo"], "includeRecentPosts": true, "maxPostsPerProfile": 6}'

JavaScript (apify-client):

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('apricot_blackberry/instagram-creator-analyzer').call({
  usernames: ['natgeo'],
  includeRecentPosts: true,
  maxPostsPerProfile: 6,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();

Python (apify_client):

from apify_client import ApifyClient
client = ApifyClient(os.environ["APIFY_TOKEN"])
run = client.actor("apricot_blackberry/instagram-creator-analyzer").call(run_input={
    "usernames": ["natgeo"],
    "includeRecentPosts": True,
    "maxPostsPerProfile": 6,
})
items = client.dataset(run["defaultDatasetId"]).list_items().items

Use from automation platforms

• n8n / Make / Zapier: use the native Apify integration and pick instagram-creator-analyzer by name.
• LangChain / LlamaIndex: use the Apify tool wrappers (ApifyActorsTool / Apify Actor loader) with the actor id above.
• Webhooks: Apify can fire a webhook when a run finishes (or on other run-lifecycle events), so you can trigger a downstream pipeline step the moment fresh creator data lands — configure one from the actor's Integrations tab or the API.

Pricing & performance

Pay-per-event, billed by the Apify platform:

Event	Price	Charged
Actor start	$0.10	Once per run (1 GB memory)
Profile analyzed	$0.08	Per creator row in the dataset

A typical profile analyzes in well under a minute; a batch of 50 creators completes in a few minutes for roughly $0.10 + 50 × $0.08 = $4.10 (~$0.082/creator). Residential proxy traffic (used only for the Instagram-direct method) is billed to your Apify account additionally. Rotation, rate limiting, and clean JSON output are included.

Instagram creator analytics use cases

• Brand managers / influencer marketers: vet creators' real follower counts, engagement, and niche before committing budget.
• Agencies: report accurate creator metrics to clients with a documented per-source audit trail (meta.methodResults).
• Talent scouts: find emerging creators by tier (nano → celebrity) and posting cadence before they get expensive.
• AI research agents: pull structured creator data on demand via MCP — audience tier, engagement rate, and sponsorship estimates in one typed row.

Works great with

• TikTok Profile Scraper — compare Instagram performance to TikTok.
• Audience Authenticity Audit — deeper follower-quality analysis.
• Creator Discovery Engine — find creators before analyzing them.

FAQ

Does it need my Instagram login? No. It never logs in and only reads public data. Private accounts return profile-level stats only (isPrivate: true).

Why is engagement 0 for some profiles? Post-level data requires the Instagram-direct method to succeed. When it doesn't, engagement fields are 0, posting.frequency is "unknown", and no engagement-based sponsorship adjustment is applied — the row tells you what it doesn't know instead of guessing.

How accurate are follower counts? Counts are merged across up to 7 sources with outlier rejection; abbreviated figures (e.g. 268M) are expanded and cross-checked against exact values when available.

Are sponsorship estimates guaranteed? They are tier-based market estimates with every adjustment documented in sponsorshipEstimate.adjustments. Use them as a negotiation baseline, not a quote.

Changelog

2026-08-14

• Output dataset schema published (.actor/dataset_schema.json) — rows are now validated and render a clean Overview table; README restructured for AI-agent consumption (MCP snippet, exact field tables, error semantics).
• Fixed abbreviated-count parsing: 30K / 268M / 1.2B style figures in Instagram og-tags, Wayback snapshots, and Social Blade pages now parse correctly; fixed a Social Blade field mix-up that could report the following count as the posts count.
• Honest sponsorship estimates: when no post data is available, the engagement-based adjustment is skipped (previously an unwarranted "low engagement −15%" was applied) and noted in adjustments.