WebmasterID docs

Start here

This is the operator manual for WebmasterID. Each page is written so a real operator, developer, or coding assistant (Claude Code, Codex) can follow it end-to-end and end up with a working install, a verified site, and a clean dashboard.

The hardest thing in analytics installation is being precise: which file, which line, which check. We try to be exact. When framework details vary, we say so and give a safe generic version plus framework-specific examples.

Pages

• Install the tracker

  Step-by-step install for static HTML, Next.js, React/Vite, Astro, WordPress. With verification + troubleshooting.

  /docs/install

• Events reference

  What page_view, outbound_click, download, cta_click, form_submit, signup, purchase, and bot_visit actually contain.

  /docs/events

• CTA tracking

  Tag CTAs with data-wmid-cta. Tag forms with data-wmid-form. Use stable, semantic ids — no PII, no keyword stuffing.

  /docs/cta-tracking

• Campaign links

  Build UTM-tagged URLs in /links. Strict utm_*-only — no gclid, no fbclid, no PII in parameters.

  /docs/campaign-links

• Verify install

  Every status the Verify Install probe returns and the operator action for each — with a single canonical table.

  /docs/verify-install

• Agent connection guide

  Tracker collects. Agent analyses. Humans approve. The boundary between Core and webmasterid-agent.

  /docs/agent

• API reference

  Tracker URL, ingest endpoint, public site_id, server-side concepts, agent endpoints, health checks. With placeholders only — never real secrets.

  /docs/api

• Privacy practices for operators

  What you should disclose, what consent-first loading looks like, what never to put in CTA ids or UTM parameters.

  /docs/privacy

Operator principles

WebmasterID is intentionally small and predictable. Three rules explain almost everything about how it behaves:

• Tracker collects. The browser tracker is a passive data-collection layer. It records page-views and a short list of declarative events, then sends them to the ingest API. It never modifies your page, never shows different content, never speaks to other users.
• Agent analyses. The agent reads sanitised analytics aggregates and (in webmasterid-agent) crawl intelligence. It produces recommendations, reports, and paste-ready prompts. It never commands the tracker, never fakes events, never auto-deploys.
• Humans approve. Every external change — repo edit, deploy, outreach, social post, ad — happens through you, never through us. Approving an action draft is operator authorisation, not auto-execution.

Related

For positioning + use cases, see /use-cases. For the system architecture, see /architecture. To request access to the private beta, see /request-access.