The cheapest fraud is the fraud you never buy. The main filters work pre-bid, so requests from bad sources are refused before any money moves. Source: Ad Fraud Protection.
What one request goes through
- A request arrives — a source offers an ad space.
- Blocklists are checked — known bad sources stop here.
- The name is checked — junk and fake domains are refused.
- Signals are checked — invalid and unmeasurable traffic.
- Clicks are scored — after the ad has shown.
- Bad sources are blocked — automatically, for next time.
What the pre-bid layer covers
- Blocklists, kept current automatically with a full history
- Junk and fake domain names (refused before bidding since v7.9)
- Invalid traffic signals
- Unmeasurable sources
Types of fraud detected
Bot traffic, invalid traffic, data center traffic, proxy and VPN abuse, click flooding, impression fraud, attribution fraud and abnormal behaviour.
Does it slow the auction?
No — pre-bid decisions take milliseconds.
Is blocked traffic billed?
The plugin FAQ states that traffic blocked as invalid is not counted or billed.
Scale
More than 410 million malicious IPs blocked and 9.5 billion auctions scanned every month.
Custom rules
Rules can be set per advertiser, campaign or traffic source, with whitelists for trusted partners.
Related: What can be blocked, and the four action levels · How AI-powered fraud detection works
Nov 13, 2026