🤖 Know Your Traffic — What's Real, What's Not

Separate the bots from the buyers. BotTollbooth brings transparent traffic intelligence to the open web. See the true split between real visitors and bots, category by category, with every answer — and the business impact — laid out honestly. Open source core, free and self-hostable. This demo runs entirely in your browser.

⚙️ Configure Your Traffic Profile

Start from a real-world traffic profile, or tune the mix yourself — how much is real, what kind of bots you're seeing, and what your ad revenue would be worth if that bot traffic were real humans.

Every answer shows its assumption — transparency is the whole point.

Human traffic share ?

55%

AI crawler weight ?

Scraper / spam weight ?

Search engine weight ?

Monitoring weight ?

Revenue per 1k views (RPM) ?

$15

Bot fill scale ?

50%

Sample size ?

2,000

Period

📊 Traffic Summary

Key numbers from your current traffic profile. Recoverable Revenue = the ad revenue being lost to bot traffic — what you'd earn if those were real humans. Every number shows its assumption, because transparency is the whole point.

📈 Traffic by Category

Who's actually visiting. Search engines and AI crawlers act differently from spam scrapers — not all bots are equal.

📋 Category Details

What each bucket means and how we identified it — no black box. You can verify every classification.

Category	Count	%	Identified by

🎯 Crawl Purpose & Value Exchange

Not all crawls are equal. AI training crawlers copy your content and send no visitors back; search crawlers pay you back with organic visits. This panel shows the true exchange rate for your traffic.

Purpose	Requests	Share of crawls	Returned to you

📦 Bandwidth & Cost Impact

Every bot request still costs you bandwidth. This is the egress + CDN bill the crawlers quietly add to your hosting, with the AI-training share broken out separately.

Metric	Bot traffic	AI-training crawlers only

🎯 Session Probe (Headless & Sensor Detection)

Server logs can't see everything. Embed a tiny probe.js in a page and it reports client-side signals — navigator.webdriver, software (SwiftShader) WebGL renderer, plugin count, and sensor APIs — so headless shells and virtualised scrapers get flagged that a UA-only scan would miss.

Signals this browser reports

Integrate — 4 lines

<script src="https://your-api/probe.js"></script>
<script>
  const s = BotTollboothProbe.collect();
  fetch('/api/v1/probe', { method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({ namespace:'mysite.com',
      userAgent: navigator.userAgent, signals: s }) });
</script>

Probe signals feed the same classifier as UA signatures — see probe:webdriver / probe:softwareRenderer in the engine.

🛡️ CoMP / EU Opt-out & Disclosure

Below the answers, the actions: which observed AI crawlers to opt out (training), which to allow (search), plus a generated robots.txt and a CoMP / EU AI-act disclosure you can publish as-is.

Crawler	Requests	Verdict	Why

robots.txt

Disclosure (CoMP / EU AI-act)

🔍 Classify Any User-Agent

Paste any user-agent string and see exactly how we classify it — with the confidence and the signal that produced the answer.