Vet creators with data, not media kits

The vetting stack in three endpoints

• channel-stats / profile — followers, bio, verification.
• channel-videos / channel-posts — the last N posts with per-post metrics.
• comments — sample real comments to spot bought engagement and brand-safety issues.

Compute a real engagement rate

const base = "https://api.scrapersocial.com/v1";
const h = { Authorization: "Bearer sk_live_..." };

const profile = await (await fetch(
  `${base}/tiktok/channel-stats?handle=${handle}`, { headers: h })).json();
const videos = await (await fetch(
  `${base}/tiktok/channel-videos?handle=${handle}&limit=20`, { headers: h })).json();

const avgEngagement = videos.data
  .map((v) => v.metrics.likes + v.metrics.comments)
  .reduce((a, b) => a + b, 0) / videos.data.length;

const rate = avgEngagement / profile.data.metrics.followers;

The pair of calls above costs 63 credits: 3 for the profile and 60 for twenty videos at 3 credits each. That's about 32 cents to vet a creator on real numbers, or a 15-creator shortlist for under $5. Drop the sample to ten videos and a full agency shortlist of 30 creators still comes in around $5.

Spotting bought engagement

Follower counts lie; comment sections don't. Pull 50 comments from a creator's last few posts (2 credits per comment on TikTok, 3 on Instagram) and look at what actual humans wrote. The patterns that should worry you: a high like-to-comment ratio with near-empty comment sections, comment text that's generic enough to paste under any video, and engagement that collapses on posts without a giveaway attached. None of this needs a scoring model. It needs the raw comments in front of an analyst, or in front of your LLM with a prompt like "flag signs of inauthentic engagement in these comments."

Cross-platform by design

The same code vets an Instagram account by swapping the platform segment. For B2B campaigns, LinkedIn endpoints cover member profiles, company pages, and post engagement. Every response is the same envelope, so your scoring model doesn't care where the creator lives.

From shortlist to report

Most teams end this pipeline in a spreadsheet or a Notion table: format=csv on the list endpoints skips the JSON-to-rows step. For recurring monitoring (tracking a signed creator's engagement over a campaign), schedule the same two calls daily and diff the metrics; cached pricing keeps repeat pulls cheap, and fresh=true is there for the days that matter.