Bot Traffic IndexThe bot problem, quantified

Bot Traffic in GA4: The Zero-Second Session Test

A store's GA4 report shows a spike in sessions from Ashburn, Virginia. The store does not market in Ashburn. It does not ship to Ashburn. The sessions have zero engagement time and one pageview each. The traffic is bots, and GA4 is the only place the store can see them, because GA4 hides the IP, the ASN, and the user agent, and leaves the store with the three signals that actually work.

The signal GA4 actually gives you

GA4 gives you a fixed set of dimensions and no IP address, no ASN, and no user agent string. Everything you can learn about bots in GA4 works inside that constraint, and the constraint is the point: the signals that survive are the ones that do not need the network. Engagement time, session counts, geography, and the tech dimensions are what remain, and they are enough, if you know what to look for. GA4 drops known bots before you see them and reports nothing about what it dropped, which means a clean report is not evidence of clean traffic. It is evidence that the known bots were filtered. The sophisticated traffic, the kind that is not on a list, is still in the numbers, and the numbers are where you find it.

The server log gap is the context: GA4 only sees clients that execute JavaScript, so a scraper pulling forty thousand product URLs never appears in analytics at all. The bots in GA4 are the ones that run JavaScript, and they are the ones that skew the metrics the business reads.

Zero-second sessions at scale

The first signal is engagement. Hundreds of sessions with zero seconds of average engagement time and one pageview each is not a user behavior. It is a bot behavior, or a tag firing before the page is interactive, or a prefetch. The boring explanations are real and they are usually the answer for a handful of sessions. At scale, hundreds of them, the boring explanations stop working, because a slow landing page does not produce hundreds of identical zero-second sessions from the same cities. The test is the pattern, not the single session: zero engagement, one pageview, repeated, from the same places, at the same hours. The fake signups post is the same signal in a different metric: a list that grows without engagement is a list full of bots, and the sessions that created it look exactly like these.

The datacenter city spike

The second signal is geography, and it is the most reliable one GA4 gives you. GA4 derives geography from the IP address at collection time and then discards the address, so traffic originating in a datacenter reports the datacenter's city. Ashburn, Boardman, Council Bluffs, The Dalles, Dublin, and Singapore in your top cities, when you do not market in those places, is almost never a boring explanation. It is datacenter traffic, and datacenter traffic in a store's analytics is bots until proven otherwise. The cities are the giveaway because they are the places the big cloud providers put their servers, and the bots rent the same servers. A spike from one of those cities, with zero engagement time, is the finding. Either signal alone is a suspicion. Both together are a conclusion.

What to do with it

Score the traffic on several independent signals and stamp the score onto the session, instead of deleting the session. A single signal is a guess. Several signals that agree are a finding, and the score is the evidence you take to the ad platform. Google gives you 60 days to ask it to investigate invalid traffic in Google Ads, and the claim needs the signals documented, not just the suspicion. The ad spend post is the cost of not doing this: bots inflate clicks, poison the optimization signal, and hand the platform a conversion history that points at the wrong people. The zero-second session is the cheapest signal in the report, and the one most teams never look at.

Frequently asked questions

Why do datacenter cities show up in GA4?

It is the most reliable tell GA4 gives you. GA4 derives geography from the IP address at collection time and then discards the address, so traffic from a datacenter reports the datacenter's city. Ashburn, Boardman, Council Bluffs, The Dalles, Dublin, and Singapore in your top cities, when you do not market there, means datacenter traffic.

Does a clean GA4 report mean there are no bots?

No, and that is the most common misreading. GA4 drops known bots before you see them and reports nothing about what it dropped. A clean GA4 report means the known bots were filtered. It says nothing about the sophisticated traffic that was not on a list.

Should I delete bot sessions from GA4?

Score the traffic on several independent signals and stamp the score onto the session, instead of deleting the session. A single signal is a guess. Several signals that agree are a finding. Deleting sessions destroys the evidence and makes the problem impossible to measure.

How long do I have to claim back invalid ad traffic?

Google gives you 60 days to ask it to investigate invalid traffic in Google Ads. The window matters, because the claim needs the data, and the data ages out. Check the traffic monthly, and file the claim inside the window, with the signals documented.