Skip to answers
Knowledge BaseTraffic quality & conversion answersTechnical documentation ↗
Browse answers

Traffic Detective

Can browser fingerprinting detect advanced automation?

In short

Browser inconsistencies can contribute evidence, but sophisticated automation can imitate normal browsers. No single fingerprint establishes bot identity.

Research context

OWASP describes fingerprints as one signal in a layered assessment, and notes their privacy implications. Browser consistency can support a hypothesis but cannot certify humanity.

Now again, that says exactly nothing.

So let's start with the basics.

People often start with user agents. Some "security solutions" rely on user agents and IP ranges alone. That's cute, but it misses a lot of automated traffic.

CreepJS and AmIUnique show how easily a person can be identified online from passively collected attributes alone.

However, both are outdated. Projects like these need constant updates for new browser versions to stay relevant. In one case, a small difference in Math can strongly indicate the visitor's operating system, but only in Chrome versions above 150.

Signals like these can be ephemeral, lasting only a browser version or two. You can't rely on Claude or ChatGPT to solve advanced automation for you.

You need a researcher who knows what they're doing.

The Fallacy of Bot Scores

Bot scores are often treated as a source of truth.

An example (it may no longer work when you read this): https://recaptcha-demo.appspot.com/recaptcha-v3-request-scores.php

For a time, reCAPTCHA scores came back as only 0.9 or 0.1, without really separating users from bots.

Scores like this one and Cloudflare's share a weakness: the majority of traffic sets what looks normal, even when the majority is wrong.

So a large enough influx of bots can tip the scales and start to look legitimate.

Next step: What if the evidence is inconclusive? · Related guide: Can bot filtering hide real customers?

Sources and further reading

Button Solutions Knowledge Base Evidence before action.