How to identify bot traffic in GA4: GA4 removes the bots it knows, so what is left to find are the unknown ones. They leave a pattern in your reports.
What Google already removes
Traffic from known bots and spiders is excluded automatically, identified “using a combination of Google research and the International Spiders and Bots List, maintained by the Interactive Advertising Bureau”. You cannot disable it or see how much was removed (Analytics Help).
Five checks
Open Reports > Acquisition > Traffic acquisition (Analytics Help). Pick a date range with the suspected spike.
| # | look at | bot sign |
|---|---|---|
| 1 | sessions by source / medium | one source with a jump in sessions and no matching ad spend or campaign |
| 2 | sessions by country | a spike from one region unlikely to read your language |
| 3 | sessions by landing page | one page taking a surge, often a page nobody links to |
| 4 | engagement rate and average engagement time | a row with many sessions and engagement near zero, or a suspiciously identical time on every session |
| 5 | key events (conversions) | a surge of sign-ups or form submissions with gibberish emails and fake names |
These match Cloudflare’s list of bot traffic hallmarks: abnormally high pageviews, abnormally high bounce rate, surprisingly high or low session duration, junk conversions, and a spike in traffic from an unexpected location (Cloudflare).
What “engaged” means: a session of at least 10 seconds, or with a key event, or with at least 2 pageviews. Bounce rate is the opposite of engagement rate (Analytics Help). Bots that load one page and leave land in the bounce rate.
Next: confirm outside GA4
A pattern in GA4 is a lead. Confirm it in your server or CDN logs: the same IP range, the same user agent, requests too fast for a person. Steps: how to check if traffic is real.
What GA4 cannot show you
- Bots that do not run JavaScript never fire the GA4 tag, so GA4 does not see them at all. They show up in server logs.
- Bots from the known list are removed before you see them. There is no count.
- Click farms use real browsers and people; the engagement can look normal. See click farm.