Glossary
What are bots in advertising?
Bots are automated software programs that simulate human activity — clicks, installs, in-app events, page views — most often used to commit advertising fraud and inflate campaign metrics.
Not all bots are fraud
A meaningful share of non-human traffic is legitimate: search engine crawlers, uptime monitors, accessibility tools, link previewers. Well-behaved ones identify themselves and respect exclusion rules.
Fraudulent bots do the opposite — they impersonate human users precisely because being counted as human is the point. The distinction in practice is between traffic that declares itself and traffic that hides.
What they simulate
- Impressions and clicks, to consume budget on inventory nobody saw.
- Installs and in-app events, to claim performance payouts.
- Session behaviour — scrolling, dwell time, navigation — to defeat engagement-based detection.
- Human-like timing, with randomised intervals rather than machine regularity.
Detection and the arms race
Detection works on the fact that simulating humanity thoroughly is expensive. Bots are given away by absent variance, impossible device configurations, data-centre network origins, event sequences that skip steps a real interface requires, and behaviour that is statistically too clean.
Every one of those signals can be addressed by a sufficiently motivated operator, which is why this is an arms race rather than a solved problem. Detection improves, sophistication improves, and the equilibrium moves.
The durable defence is economic rather than technical: make the payout conditional on something expensive to fake. A bot can generate a click for nothing and a genuine subscription for very little less than the subscription is worth.
Bots and AI surfaces
Automated traffic is a real concern for AI publishers, and the shape differs. There is little value in faking a click that does not exist; the exposure is manufactured conversational volume — automated traffic inflating a publisher's apparent inventory and its apparent intent quality.
That is a supply integrity problem for the publisher and the platform rather than a click-fraud problem for the buyer, which is another reason knowing which publishers you are actually buying matters on this surface.
Common questions
How much traffic is bots?
Estimates for overall internet traffic run high and vary wildly by methodology and by whether legitimate crawlers are included. The number that matters is the share in your own paid traffic, which is measurable by source.
Do CAPTCHAs stop ad fraud bots?
Not usefully. Ad fraud bots operate on impressions, clicks and SDK traffic, none of which pass through a CAPTCHA. Detection has to happen at the measurement layer.
Where we write about bots
More in fraud and validation
How invalid traffic is manufactured, and how it is caught.