> ## Documentation Index > Fetch the complete documentation index at: https://docs.zeroclick.ai/llms.txt > Use this file to discover all available pages before exploring further. # What is Page views? > Page views reports every page-content request your marketing site serves so you can see who reads it - agents, crawlers, and people - and how that traffic converts. What it captures, how viewers are classified, the privacy model, and the API key scope. Page views tells you who reads your marketing site and how that top-of-funnel traffic turns into agent purchases. Your site reports each page-content request to ZeroClick, which classifies the visitor against the same agent registry your storefront uses, so a page view of your pricing page and a later purchase through your storefront line up as one journey. It is a separate capability from [Agentify](/agentify/overview), enabled on its own. Agentify changes what an agent *receives* (markdown instead of HTML); Page views only *observes* the request and never changes the response. Turn on either, both, or neither. Report page views one of two ways: mount the [SDK middleware](/page-views/sdk-middleware) in your site so every page request is reported automatically, or send each view to the [REST endpoint](/page-views/rest) from any stack. ## What is reported A page view is one `GET` request to a page-content path. Requests for assets — anything whose final path segment has a file extension, like `/_astro/app.css`, `/favicon.ico`, or `/logo.png` — are skipped, as are redirects. Everything a reader actually loads to view a page is reported: your pages for people, and the same pages for agents and crawlers. There is no user-agent gate at capture. A crawler, an unknown agent, and a browser are all reported and told apart afterward, so a new agent shows up in your numbers without any change on your side. ## How a viewer is classified The visitor's user-agent, `Accept` header, and IP are read to classify the request, then discarded. What is stored is coarse and non-identifying: * **Client kind** — `agent`, `browser`, or `other`. * **The agent**, when one is recognized — its registry slug (`gptbot`, `perplexitybot`, …) and operator (OpenAI, Perplexity, …), plus a coarse agent type. * **A user-agent family** and the registry version that classified it, so history can be re-bucketed when the registry improves. No raw user-agent or IP is ever stored. Each view carries the same anonymous viewer id your storefront traffic already does — a hash of the visitor's IP and user-agent — so a page view, a later storefront visit, and a purchase from the same client line up as one journey, without either side ever holding an address. Two coarse context signals ride along when the edge exposes them: the visitor's **country** (from Cloudflare's `cf-ipcountry`, clamped to a 2-letter code) and the **referrer host** — the referring domain only, never the full referrer URL, so a referrer's own path or query never leaves the visitor's site. Both are optional and hold nothing that identifies a person. ## Two timestamps Each view carries two clocks, so you can measure delivery lag and trust the right one: * **`captured_at`** — when your edge observed the request. Present only when your reporter sends it; it is never invented. * **`received_at`** — when ZeroClick received the beacon. This is the authoritative time. ## The Page views key scope Reporting authenticates with an API key carrying the `page-views:write` scope. Mint one in the [dashboard](https://dashboard.zeroclick.io) under **Settings → API keys**: create a key and enable the **Page views** group's **Write** toggle. The scope grants `POST /v1/page-views` and nothing else, so the key can live on your marketing site's edge without exposing your catalog, usage, or Agentify APIs. A key can carry both `page-views:write` and `agentify:convert` when you run both from the same edge, or just one. ## Next steps Report every page view from your own site, fire-and-forget. Send a page view from any stack with one POST.