What percentage of website traffic is bots?

Industry measurements put automated traffic between a third and half of all web requests, but the share on any one site depends entirely on where its traffic comes from. A site buying cheap clicks can run far past half; a site with organic search readers can sit under ten percent.

Does Google Analytics show bot traffic?

GA4 filters known declared bots automatically but publishes no report of what it removed, and it does not classify what it kept. That gap is what traffic verification fills: every visit is classified as human, suspected bot, or confirmed bot, with the reason stored.

What is the fastest single check?

Open the realtime view and watch who is reading right now. A wave of visits with no mouse movement, no scrolling, and identical timing reads as automation within minutes, long before a daily report exists.

How to tell if your website traffic is bots

Something about the numbers feels wrong, but a feeling is not a finding. These are the five signs worth checking, ranked by how much each one actually proves, and the checks that turn a hunch into evidence.

Last updated Aug 29, 2026

The five signs worth checking

A bounce rate near 100 percent on one source while the rest of the site sits at its usual level.
Sessions under five seconds: nobody read anything, thousands of times in a row.
One page per session across an entire source, when your navigation normally pulls people deeper.
Geographic concentration that makes no sense for the campaign, like one city carrying half a country.
A desktop heavy mix on traffic that should be mobile, because cheap automation runs on servers, and servers report themselves as desktops.

One sign is a hint, several are a pattern

Every sign on that list has an innocent explanation on its own. A landing page with no internal links produces one page per session honestly. A viral post in one city produces geographic concentration honestly. What has no innocent explanation is uniformity: real audiences are messy, and a source where every visit behaves identically is describing a script, not a crowd.

Turn the hunch into evidence

A hunch becomes a finding when a visit fails a check it should pass. The four layers of detection verify rather than guess: a claimed crawler is checked against published addresses, a data center range match is read as a first class signal, and a client that never ran our JavaScript is marked with exactly that reason. Every verdict keeps the reason it was reached, so you can read why a visit was flagged instead of trusting a percentage.

Check the traffic your tracker never saw

AI crawlers and most scrapers never execute JavaScript, so a browser tracker cannot see them at all. Server-side collection exists for exactly this blind spot, and comparing the two counts is itself a bot check: requests the tracker never saw are clients that never ran your page.

Then act at the source, not the site

A verdict without an address is trivia. Segment the flagged traffic by source, placement, and sub ID, then work through which sources to cut first with the ranking that starts from wasted spend rather than from the scariest percentage.

Frequently asked questions

What percentage of website traffic is bots?: Industry measurements put automated traffic between a third and half of all web requests, but the share on any one site depends entirely on where its traffic comes from. A site buying cheap clicks can run far past half; a site with organic search readers can sit under ten percent.
Does Google Analytics show bot traffic?: GA4 filters known declared bots automatically but publishes no report of what it removed, and it does not classify what it kept. That gap is what traffic verification fills: every visit is classified as human, suspected bot, or confirmed bot, with the reason stored.
What is the fastest single check?: Open the realtime view and watch who is reading right now. A wave of visits with no mouse movement, no scrolling, and identical timing reads as automation within minutes, long before a daily report exists.