Going further
Bots & AI answers
Automated traffic now makes up a serious share of hits on content-heavy sites. NumberHill classifies it at the ingest layer — before anything is written to your analytics — so it never inflates visitor counts or dilutes revenue per visitor.
How it works
Every request to /api/collect has its user agent matched against a signature list. A hit records to a separate bot table and returns { ok: true, filtered: "bot" }; it never lands in events. That keeps the bot report accurate while leaving your real metrics clean.
What gets detected
| Category | Examples | Counted as a visitor? |
|---|---|---|
| AI crawler | GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Bytespider and ~45 more | No |
| Search engine | Googlebot, Bingbot, DuckDuckBot, Baiduspider, YandexBot, Applebot | No |
| Scraper | AhrefsBot, SemrushBot and similar SEO crawlers | No |
| Monitoring | UptimeRobot, Pingdom and uptime checks | No |
| Headless | Headless Chrome, Puppeteer, Playwright, curl, wget | No |
The full signature list is public. The AI crawler directory has a page per crawler with its operator, robots.txt token and verification method, the user-agent checker classifies any log line with this same code, and the robots.txt generator builds block rules from it.
Filtered ≠ blocked
NumberHill never blocks anything. It ships norobots.txt rules, sets no rate limits, and answers 202 to every agent — filtering happens only in your analytics. Blocking a crawler is your CDN's or robots.txt's job, and for AI crawlers it is usually the wrong move: an engine that cannot read a page cannot cite it.AI visibility: visitors AI answers send
When ChatGPT, Perplexity, Claude or Gemini cites your site and someone follows the link, that visit arrives with the engine's referrer. NumberHill attributes it to the engine by name and carries it through to conversion and revenue, like any other channel. This is measured in the browser, so it works on every platform — Shopify and other hosted storefronts included. The AI visibility report shows:
- Which engines send visitors, and how many of them buy.
- Which pages the citations point at — where those visitors land.
- Who bought — the individual customers an AI answer produced, with what they spent.
Two undercounts to know about: AI desktop and in-app browsers often send no referrer, and those visits land in Direct. Google AI Overviews and inline Copilot answers refer as plain google.com / bing.com, so they count as search.
Why this matters for revenue
Revenue per visitor is a ratio. Inflate the denominator with crawler traffic and every channel looks worse than it is — unevenly, since crawlers hit blog posts far harder than pricing pages. Filtering at ingest is what keeps the comparison between channels honest.