<b>Bot and junk traffic — the filter rule stack</b>

<b>Bot and junk traffic — the filter rule stack</b>

Unfiltered bots poison your optimization data, not just your stats. The algo learns from garbage. Filter at the door.

Layer the filters in this order (cheapest check first):
— 1. IP reputation / known datacenter ranges. Block AWS, GCP, OVH IP blocks for consumer offers — real users aren't in datacenters. Cheapest, catches the most.
— 2. User-agent sanity. Block empty UAs, headless markers, and outdated browsers that scream automation.
— 3. Click timing. Multiple clicks from one IP within seconds = a crawler or a competitor. Rate-limit per IP.
— 4. Honeypot link. A hidden link real users never see. Anything clicking it is a bot — auto-blacklist that IP/placement.
— 5. Click-to-LP-load ratio. Clicks that never fire the LP-loaded pixel are likely scanners.

Route decisions:
— Don't send filtered traffic to the offer (you'll pay for it or burn caps). Send to a blank 200 page so the source doesn't flag the link as broken.
— Log filtered clicks separately so you can prove fraud rates to your source rep.

Review the filtered-traffic log weekly to blacklist recurring bad placements.

Run this before every launch.