Creator Rate-Card Pricing Benchmark

Real, live-fetched follower data — not manually typed numbers — run through a transparent benchmark formula. The existing Apify Store competitor in this space is a pure calculator: its own README states it runs "without third-party API keys, scraping, or private profile access," meaning you have to already know (and type in) the creator's follower count and engagement rate yourself. This actor fetches that data live from the creator's real public profile instead.

A finding worth being direct about

The build PRD's own stated goal was "observed pricing" — real, itemized deal prices — not a calculator. Before writing any scraper code, three candidate sourcing paths were tested live:

1. Public rate cards individual creators publish themselves — real, but scattered across personal sites and media kits with no directory or discoverable URL pattern. Not viable for a general "look up any creator" tool.
2. Marketplace-disclosed pricing — tested four real creator marketplaces live. Collabstr has real per-creator rate data, but the entire site (even robots.txt) sits behind a Cloudflare managed JS challenge. Shoutcart returned a flat 403. influence.co requires a full Google OAuth login just to browse. Ainfluencer's public browse page loads but shows zero visible price data — rates are gated behind a contact/signup flow. Dead end across every real marketplace tested.
3. Real, live public follower/engagement data (the PRD's own suggested fallback) — confirmed technically working on all three platforms, see below.

Per the PRD's own explicit fallback instruction, this actor is built as a benchmark calculator informed by real, live-fetched public data, not observed deal prices — which are confirmed inaccessible, not merely inconvenient. Every record honestly reports dataSource: "inferred_benchmark", since the other two paths didn't pan out.

What's actually verified, per platform

Platform	Follower data	Engagement signal	Confidence cap
TikTok	✅ Real, live, no login — followerCount from the embedded #__UNIVERSAL_DATA_FOR_REHYDRATION__ JSON	✅ Real — average likes-per-video-ever ÷ followers (see caveat below)	high
YouTube	✅ Real, live, no login — subscriberCountText from the channel's /about page	❌ Not fetched in this build	medium
Instagram	⚠️ Real when it works, but confirmed IP-dependent (see below) — parsed from the public og:description meta tag	❌ Not available without login	medium

None of these require logging in. dataConfidence is high only when both a real follower count and a real engagement signal were fetched (currently TikTok only); medium when only follower count was fetched; low when the live fetch failed and the run fell back to a manually-supplied followerCount input instead.

Instagram specifically was confirmed working during local development, then confirmed failing consistently once deployed and run on the real Apify platform — the same code, same handle, different result depending on where it runs. The real cause, found via diagnostic logging on a live platform run rather than guessed: Instagram returns a genuine HTTP-level failure (net::ERR_HTTP_RESPONSE_CODE_FAILURE) specifically to requests from Apify's cloud IP range, while the identical request succeeds from a non-cloud IP. This is a real, external IP-based block — not a bug in this actor, and not something a residential-proxy-free fix can solve. TikTok and YouTube were both tested from the same Apify cloud run and worked without issue, so this is Instagram-specific, not a platform-wide problem. A residential proxy would likely resolve it; that's out of scope for this build. dataSource/dataConfidence on a failed Instagram lookup are reported honestly as failed/low rather than silently retried into something that looks like success.

Three things found while building, not assumed

• Instagram needs a real browser; TikTok and YouTube don't, technically. A plain curl request to an Instagram profile (even with a real browser User-Agent) gets an empty, generic-titled, login-walled shell with zero data. A real Playwright browser gets past this and the og:description meta tag reliably has real follower/following/post counts. TikTok and YouTube's data is present in raw HTML even via plain curl — Playwright is used for all three anyway, for one consistent code path rather than mixing HTTP clients.
• YouTube's /about page routes through a real interactive cookie-consent page in a browser, not just a redirect. curl -L auto-follows straight to the real content, landing on a URL with ?cbrd=1&ucbcb=1 appended. A real Playwright browser instead gets shown consent.youtube.com's actual interactive consent UI — confirmed by inspecting the captured HTML directly (a real, separate document, not an empty stub). Appending those same query params to the initial request URL skips the interstitial in Playwright too.
• TikTok's public profile page has no per-video breakdown, only lifetime cumulative stats. heart (total likes ever) and videoCount are both real and public, so heart ÷ videoCount ÷ followerCount gives a genuine average-engagement proxy — but it's a lifetime average, not a "last 10 posts" engagement rate. Documented here rather than presented as more granular than it is. Also found live: TikTok's own stats.heartCount field returned a nonsensical negative number on a real profile (an apparent counter-overflow bug on TikTok's end) while the newer statsV2.heartCount matched heart exactly — statsV2 is used for this reason.

Modes

Mode	Input	Output
creator_lookup	creatorHandle (+ optional followerCount fallback, niche, geography)	One benchmark result for that creator
niche_benchmark	niche / geography (no handle)	One result per follower tier (nano/micro/mid/macro/mega), using a representative follower count per tier

The pricing formula — transparent, not a black box

Base rate: $100 per 10,000 followers, a widely-cited creator-economy rule of thumb (cross-checked during research against real published 2026 benchmark ranges — nano ~$25-$500, macro ~$10,000-$50,000+ — and lands within them). Then adjusted by:

• Platform — YouTube ×3.0 (a dedicated integration is real production effort, not a single post; Instagram/TikTok both ×1.0)
• Niche — a modest table (finance/B2B/SaaS ×1.8, beauty/fashion/luxury ×1.3, tech ×1.1, entertainment/comedy ×0.8, unrecognized or blank niches ×1.0 — no guessed penalty)
• Geography — a short list of premium ad markets (US/UK/Canada/Australia/NZ + several Western European countries) gets ×1.3; anything else, including blank, is ×1.0 — again no guessed penalty for unrecognized geographies
• Engagement — only applied when a real signal was fetched (TikTok); scales the estimate relative to what's typical for that creator's own follower tier, since smaller accounts genuinely have higher typical engagement rates than mega accounts, a well-known real industry pattern

Lower confidence widens the output range rather than pretending to more precision than the data supports: high confidence → ±15-20%, medium → ±30-40%, low → ±50-80%.

Output schema

{
  "creatorHandle": "string",
  "platform": "instagram | tiktok | youtube",
  "followerTier": "nano | micro | mid | macro | mega",
  "followerCount": "number | null",
  "estimatedRateRange": { "low": "number", "high": "number" } ,
  "currency": "USD",
  "dataConfidence": "high | medium | low",
  "dataSource": "inferred_benchmark",
  "niche": "string | null",
  "geography": "string | null",
  "checkedAt": "ISO timestamp",
  "status": "success | failed",
  "errorMessage": "string | null"
}

currency is always USD regardless of geography — no FX conversion is attempted, to avoid adding staleness risk for a figure that's already a directional estimate.

Pricing

Event	Price
Run started	£0.05
Lookup success	£0.06
Lookup, no data found	£0.02
Lookup failed	free

niche_benchmark mode charges the success event once per tier row produced (5 per run).

Setup note

Playwright/Chrome base image (apify/actor-node-playwright-chrome:20) — required for Instagram specifically (see above); used for all three platforms for one consistent code path.

n8n integration

• Workflow A (trigger): on-demand lookup triggered from a brand's creator shortlist.
• Workflow B (processing): side-by-side benchmark report comparing a shortlist of creators by estimated rate and follower tier — same alerting/reporting pattern as other actors in this portfolio.