Closed-loop ad optimization

Stop paying for clicks
that never convert.

Detect & Refine watches every visitor — scroll depth, mouse velocity, rage clicks, dwell, exit intent — and grades the click quality from A+ to F. Then it tells your ad networks who to bid on, who to nurture, and who to stop showing ads to entirely.

Adaptive Siphon · Live
Good
0
Nurture
0
Dead
0
Excluded
0
How it works

Three tiny moving parts.
One closed loop.

01
Detect

A 4 KB script captures the click ID (gclid, fbclid, li_fat_id, ttclid, twclid) and silently records 30+ behavioral signals — scroll, dwell, rage clicks, dead clicks, form focus, copy-paste — and ships them via sendBeacon.

02
Grade

Every session gets a rule-based score from 0-100, then refined by Groq llama-3.1-8b. Output is a letter grade (A+ → F, X for excluded) plus an action: sale, nurture, dead, exclude.

03
Refine

Grades stream back to Google Ads, Meta, LinkedIn, TikTok, and X via offline conversion APIs. The networks learn who to bid on — and who to stop showing ads to. The loop closes itself.

Install

One script tag.
No SDK. No npm.

Drop this on any landing page where ad traffic lands. The script is non-blocking, async, and ships its own Beacon — never affects Core Web Vitals.

index.html
<script
  src="https://detectandrefine.0ncore.com/cro9.js"
  data-endpoint="https://detectandrefine.0ncore.com/api/track"
  async
></script>
The script auto-extracts gclid, fbclid, li_fat_id, ttclid, and twclid from the URL, persists a session cookie, and starts pinging behavioral metrics. No config needed.