Traffic and analytics

Last verified

Every site comes with traffic analytics. Nothing is added to your pages. We count the requests on our own servers, from your site's web server access log, so there is no tracking script and no third party involved.

Open the analytics

  1. Open Sites → your site → Analytics.
  2. Choose a range: 24 hours, 7 days or 30 days. Times are in your time zone, and the header shows when the figures were last updated.

If the tab says No visits recorded yet, counting starts with the first request and the page fills in by itself. If it says Traffic analytics are not collected for this site, the site does not run in its own container with us (for example, a site we adopted), so there is nothing to count.

What you see

At the top:

  • Requests: every request, split into visitors and bots.
  • Unique visitors: the number of distinct IP addresses.
  • Response data: how much data your site sent.
  • Server errors: 5xx responses, your site failing.
  • Not found / refused: 4xx responses. Most come from scanners looking for files like /.env and bots posting to forms. Turning them away is your site working, so they are counted here and not as errors.
  • Blocked at the edge: requests aimed at your site that never reached it: known exploit probes, and requests blocked by a virtual patch (WAF). This tile only appears once there is data for it.

Below them:

  • Requests over time: a bar chart of visitors and bots.
  • Response time: the Average, Median (p50) and Slowest 5% (p95), measured on our server for each request.
  • Cache: the share of cacheable requests served from cache without running your app, split into From cache, Miss, Bypassed (for example logged-in visitors, or a query string) and Dynamic (never cached).
  • Response codes: 2xx, 3xx, 4xx and 5xx.
  • Visitors vs. bots: search engines, monitors and scrapers are counted apart from your visitors. Traffic that claims to be a browser but behaves like automation is counted as automation, and Why traffic was counted as automation lists the reasons.
  • Top pages, Top referrers, Top countries, Top browsers and Search & AI crawlers.
  • Refused requests (4xx): what was turned away and where it came from.

Under the analytics, most sites also have a Resources section that shows what the site's container used: CPU, memory, disk and network.

Privacy

  • The figures come from your site's own access log, counted on our servers. We add no script to your pages and send nothing to a third party.
  • Unique visitors counts distinct IP addresses. It is not a count of people: several people behind one office connection count once, and one person on a phone and a laptop counts twice.
  • Top countries comes from looking up each request's IP address in a geolocation database. Countries with only a couple of requests are folded into Other. Countries are kept as a count per country, never per visitor.
  • The Analytics tab shows only totals and top lists. The individual requests are in Logs → Visits (access log). See Reading your site logs.

Whether you need to mention this in your own privacy notice depends on your situation. We can't advise you on that.

From the API or an AI assistant

  • GET /api/v1/sites/SITE_SLUG/analytics returns the same figures: 24-hour, 7-day and 30-day totals, plus hourly and daily series and edge_blocked. The 4xx responses are reported apart from the server errors. sampled is false until the first count has run.
  • GET /api/v1/sites/SITE_SLUG/traffic returns disk use and network traffic. The period_* fields cover the current calendar month and the net_* fields are running totals since the last restart. They are null until the first measurement.

Both need only the read scope. An assistant uses get_site_analytics and get_site_traffic. See the API and MCP tools.

Stuck? Contact support — replies within one business day.