Real Audience, Or All Bots?

"They want $4,000 for one post. Is anyone actually going to see it?"

Follower count tells you nothing. A bought audience and a real one look identical on the profile, and by the time you find out which you bought, the money has gone.

You get one number out of 100, and every piece of it is something we measured: how the followers arrived and whether the curve has jumps no real account produces, what the engagement rate is against creators the same size, how much conversation sits under the likes, and how steady it is post to post. Bought engagement is suspiciously even; real audiences are messy.

What comes back reads like this:

"@creator scores 34/100. They gained 40,000 followers in three days in March and their comment-to-like ratio is a tenth of what creators their size run at. The likes are there. The people are not."

Every part of that score says what it was built from, so you can defend the call to a client.

⬇️ Input

The actor accepts an array of creator canonical handles, maximum creator count, and an optional proxy configuration.

Field	Type	Description
handles	Array	The creators you want checked - one handle per line.
maxCreators	Integer	Maximum number of creator profiles to process in a single run.
proxyConfiguration	Object	Optional proxy settings to rotate exit IPs per request.

⬆️ Output

Every processed creator profile outputs a structured record to the default dataset.

• canonicalHandle (Stable per-row identifier)
• credibilityScore (Composite score 0-100)
• Derived answer metrics and profile attributes.

How it works

1. Multi-Metric Extraction: Collects profile metrics, video counts, like counts, and comment threads without mandatory proxy spend.
2. Scoring: Weighs engagement ratios, repeat-commenter overlap and the shape of the growth curve into one number you can defend.
3. Ledger History Trail: Tracks history depth via daysTracked, firstSeenAt, and lastSeenAt so agencies can defend numbers in client decks.

❓ FAQ

How is the score worked out?

The score combines peer-adjusted engagement multipliers, comment author repeat rates (pod signals), follower growth continuity, and geographic commenter alignment into a 0-100 index.

Why does followerSpikeFlag return null on initial runs?

Step-change spike detection requires observed historical runs over time. On the first run, the actor returns null along with a computedStatus object noting that tracking begins today.

💬 Your feedback

Want different weights, or the data wired into something you already run? reapxdev@proton.me.

---

Unofficial - not affiliated with social networks. Collects only data visible without logging in. reapx · reapx.dev · reapxdev@proton.me