Documentation analytics: find the pages that help and the ones that fail

Docs are both support and growth: a good quickstart converts developers, a missing page creates tickets. Docs teams need signals that point at specific pages — this one is read to the end, this one has people bouncing to search, this old URL is a 404 from a blog post.

VisitTrack measures active reading time and scroll depth for every page, records 404s, rage clicks and JavaScript errors automatically, shows Search Console queries per page, and reports AI crawler visits (GPTBot, ClaudeBot, PerplexityBot and others) separately from humans. Docs on a subdomain share visitors with the main site, so docs-to-signup journeys are visible.

The documentation metrics that matter

1. 1.Engaged reading per page

   Active time (visible tab only) and scroll depth per page in the Pages tab. A reference page read for 4 minutes is doing its job; a tutorial abandoned at 20% scroll isn't.

2. 2.Failed searches

   Searches that return nothing, sent as an event. The most direct list of content to write next.

3. 3.404s and broken inbound links

   page_not_found fires on pages whose title contains "404" or "not found", or that carry <meta name="visitrack:404">. After a restructure, it shows which old URLs need redirects.

4. 4.Docs-to-signup and docs-to-activation

   Which docs pages start journeys that end in a signup or an API key. For developer products, the quickstart is often the best landing page.

5. 5.AI crawler reads and AI referrals

   Which docs pages GPTBot, ClaudeBot and PerplexityBot fetch, and how many developers arrive from ChatGPT, Claude or Perplexity. Pair with an llms.txt.

Questions docs teams ask

Which pages should we rewrite first?

High-traffic pages with low scroll depth or short active time, especially if they're entry pages from search.

What are people searching for and not finding?

Count docs_search_empty events to see how often it happens and on which pages. The dashboard counts events by name, so keep the full list of failed terms in your search backend's logs; the event's prop is handy for spot checks through the events API.

Do docs readers become users?

Filter signups by entry page, or build a funnel from a docs page to signup.

Is the copy button on code blocks working?

Rage clicks are recorded automatically, with the element's label — a cluster on a code block means something isn't responding.

How to set up docs analytics with VisitTrack

1. 1

   Add the script to your docs framework

   Any framework that lets you add a script to the head works — Docusaurus (headTags or scripts in the config), VitePress (head in the config), Nextra (the Next.js layout — see the Next.js guide). Use the same data-site as the main site.

   <script defer data-site="SITE_ID" src="https://visitrack.app/tracker.js"></script>

2. 2

   Mark your 404 page

   If the 404 page's title doesn't say "404" or "not found", add the meta tag so misses are counted.

   <meta name="visitrack:404" />

3. 3

   Track failed searches

   From your search component, send an event when a query returns no results.

   if (results.length === 0) {
     window.visitrack?.("docs_search_empty", { q: query.slice(0, 80) });
   }

4. 4

   Connect Search Console and AI crawler tracking

   Search Console in Pages & Search; AI crawler tracking needs a few lines in your server or edge config — see AI crawler tracking.

5. 5

   Count completed tutorials

   Add data-vt-scroll="tutorial_done" to the last section of each tutorial; it fires once when half visible.

Example: an API product's docs in one month

Docs on docs.example.com, ~30,000 monthly readers, same site id as the marketing site.

Illustrative numbers for a typical site of this kind, not real customer data.

The month's docs backlog writes itself: fix the webhooks guide where readers stall, write rate-limit and pagination pages, redirect the old /v1/ URLs. And since AI crawlers read the API reference daily, keeping it accurate is now an acquisition task as well as a support one.

VisitTrack vs docs-platform analytics and GA4

Third-party details as of October 2026; hosted docs platforms vary.

If your docs platform already reports search analytics, keep it for that. VisitTrack adds the cross-site view — docs to signup — plus AI crawler visibility. See the developers guide and VisitTrack vs Plausible.

What VisitTrack costs for a documentation site

Docs readers browse 3–5 pages per visit; AI crawler hits are never billed.

Every feature is on every plan; the price only follows monthly events (pageviews, custom events and payments, counted across all your sites). Bot and AI-crawler hits and session replays are never billed. Yearly billing is ten months' price for twelve, every plan starts with a 14-day free trial with no card, and there's a one-time lifetime deal from $59.