Sort traffic acquisition
Reports → Acquisition → Traffic acquisition. Sort by sessions and flag referrals from domains you don't recognise.
GA4 filters the bots it already knows about — and quietly counts the rest as real visitors. Here's what slips through, how to spot it, and how to see your true number.
At a glance
Only partly. GA4 automatically excludes traffic from known bots and spiders on the IAB/ABC International Spiders and Bots List. It does not catch sophisticated invalid traffic that mimics human behaviour, rotates IPs or runs through residential proxies — so the bot share you see in GA4 is an under-count.
For context on how big the gap can be: across 2 billion clicks in the Opticks Ad Fraud Report 2025, invalid traffic ranged from 2.18% on search to 15.9% on native ads, and the large majority of detected bot activity was bad-bot traffic with no genuine intent.
The gap
Google Analytics was built to measure human behaviour, not to police fraud. Its bot filtering is a single on/off setting that matches a known list. Anything not on that list, and anything designed to look human, sails straight through.
Ghost traffic is fake traffic that never actually visits your site. It's sent straight to Google Analytics through the Measurement Protocol using spoofed tracking IDs, inflating your sessions and referral reports without a single real pageview. Because the hit never touches your server, server-side filters can't see it.
Referral spam is the broader family: junk referrals from domains you've never heard of, designed to get you to visit them. Self-referrals, where your own domain shows up as a traffic source, usually point to a tracking misconfiguration rather than fraud — but they distort the same reports.
The common thread: all three inflate the numbers you use to make decisions, and GA4's default filtering catches almost none of it.
Step by step
You can't get an exact figure inside GA4, but these five checks surface the patterns that betray bots, ghost traffic and referral spam.
Reports → Acquisition → Traffic acquisition. Sort by sessions and flag referrals from domains you don't recognise.
Add Average engagement time. High-volume sources with near-zero engagement are almost always automated.
Traffic concentrated in one city or a single data-centre ISP is rarely a room full of real customers.
Sudden jumps in Direct / (none) with no campaign behind them are a classic ghost-traffic signature.
A source sending hundreds of sessions and zero conversions is invalid traffic, not an audience.
GA4 excludes known bots automatically — it's always on and can't be switched off. It only catches bots on the IAB/ABC list, so treat it as a floor, not the answer.
Why it matters
When bot traffic is counted as real, every metric downstream is wrong: inflated sessions, distorted bounce and conversion rates, and a cost-per-acquisition that looks better than it is. On paid channels the damage is direct — budget is spent on clicks that can never convert.
How much depends on where your traffic comes from. These are the benchmark invalid-traffic rates by channel from the Ad Fraud Report 2025:
| Channel | Invalid traffic rate |
|---|---|
| Native advertising | 15.9% |
| Programmatic / display | 15.43% |
| Social | 10.61% |
| Affiliate | 9.09% |
| SEM / Search | 2.18% |
GA4 will never show you these by channel, because it can't tell a clean click from an invalid one once the visit is recorded. A detection tag can — it validates each visit in real time against device, network and behavioural signals GA4 never collects.
Start with a free benchmark in the bot traffic checker, then deploy a detection tag to measure your real invalid-traffic rate by channel. Setup takes 5 minutes.