The challenge
Clean Skin Club sells skincare on TikTok Shop, and creators request free samples every day. Every request raises the same three questions:
- Numbers: has this creator sold enough, and posted often enough, to earn this product?
- Brand fit: do their videos look and sound like Clean Skin Club?
- Decision: who makes the call, and how fast does it reach the creator?
Each request meant a person opening the creator’s profile, checking their sales, scrolling their videos and deciding. That added up to an estimated 40 hours a week. The brand is high stakes, so the bar could not drop, but it also could not be so strict that good creators were turned away. I built the review around the content criteria the team already used and tuned it against their own past calls.
What I built
A numbers check on every request
The gap: every request needed someone to look up the creator’s sales and posting history.
- Creators who rarely post after getting a sample are turned down.
- Creators who sampled Clean Skin Club before and posted are approved again.
- Each product is set to a tier with a sales bar. Creators who sell above it are approved straight away.
- Everyone else goes to a video review, so smaller creators still get a fair look.
Now: every request is checked by the same written rules, in seconds.
AI watches the creator’s videos
The gap: brand fit was a judgment call made by scrolling a profile.
- For every creator who needs a review, AI watches four of their newest videos.
- Each video is scored on seven criteria: visual quality, aspirational look, consistency, an authentic voice, presence on camera, product demonstration and selling skills.
- Hard noes are written down: AI-generated content, heavy filters, private accounts and off-brand videos.
- Each creator is scored three times and the scores are averaged, so one odd read does not decide a sample.
Now: every creator who gets a video review is judged against the same written standard the team signed off on.
The decision lands in Cruva on its own
The gap: a decision only counted once someone clicked approve or ignore in Cruva.
- Every weekday morning the queue is reviewed before the team logs on.
- Requests for products outside the tiers are left for the team and flagged.
- Each request is approved or ignored directly in Cruva, and the reason for each call is logged.
- A one-line count of what was approved and ignored posts to the team’s Slack channel.
Now: the team no longer reviews sample requests one by one, an estimated 40 hours a week back.
What this means for your brand
If your team spends its week deciding who gets a sample, I can put the same system in place for you:
- A numbers check on every request, set by product
- AI that watches new creators’ videos and scores them against your brand
- Approve or ignore written straight to Cruva every weekday
- A weekday count in Slack so you can see what went out
I build it with your team so you own and run it after I leave.