Ad Fraud Detection
Top 10 Signs Your Campaign Traffic Isn't Human
Ten red flags hiding in your reports. Spot two or more together and it's time to verify.
By Will Harnden ·
Bots leave tells. Here are ten of the clearest, straight from the metrics you already track.
1. 100% bounce with zero-second sessions
Real people linger, even briefly. Instant bounces at scale are automated hits, not curious visitors who changed their mind.
2. Traffic spikes at 3am
Human demand follows the clock; round-the-clock or overnight surges usually mean a script running on a schedule rather than an audience.
3. Impossible click-through rates
A CTR many times the channel norm isn’t a great creative — it’s a bot clicking everything it’s served.
4. Ancient operating systems
Clusters on long-obsolete OS versions usually mean emulated or spoofed devices pretending to be real handsets.
5. Data-centre IP addresses
Consumers don’t browse from server farms. Data-centre traffic is non-human until proven otherwise, and it’s easy to check.
6. Identical user agents at scale
Thousands of “users” sharing the exact same device fingerprint is a bot signature, not a coincidence.
7. Geo that doesn’t match your targeting
Impressions from regions you excluded signal spoofing or corrupted location data leaking budget abroad.
8. A conversion cliff
Big impressions and clicks, then almost no downstream action — the funnel collapses because there was never a human in it.
9. Duplicate device IDs
The same advertising IDs appearing again and again inflate “unique reach” with repeats, recycled handsets and machines.
10. Suspiciously perfect viewability
Flawless viewability with no variance is often ad stacking or automated rendering, not attentive humans watching.
What to do when the flags cluster
One flag can be innocent. Several together means a chunk of your budget is reaching machines — and only device-level verification can tell you how much.
Treat these as a dashboard, not a tick-box checklist. Any single flag has an innocent explanation, but they rarely travel alone — and clusters of them concentrated in particular placements or sources are how you find the inventory quietly draining a campaign. When you see the pattern repeat, don’t just optimise around it: verify at the device level and cut it off at the source, because the same junk inventory will otherwise resurface under a different name next month.