How it works

Six steps, in order. Most confusion about the product comes from a step being skipped rather than failing, so it is worth knowing which one you are on.

1. Discover

We read your sitemap and follow internal links to collect the URLs your site has. Discovery lists them; it does not fetch each page. A large discovered number is not progress on its own.

2. Map to categories

You assign URLs to categories, either with a filter pattern or by hand. A category does two jobs: it is what gets a shared metadata template, and it is what selects pages for crawling.

This is the step most often missing when a site looks stuck. Category suggestions are derived from URL paths, not from crawled content, so a site that has never been crawled can be mapped on its first day.

3. Crawl

Fetches each mapped URL and stores what it found. Everything downstream reads crawl data — the score, the audit, keyword inference and link analysis — so an unmapped page is invisible to all of them.

There are two kinds. A metadata crawl reads the head: title, description, canonical, robots, structured data. A content crawl also extracts the body: headings, text, images and internal links, which is what keyword and link analysis need.

4. Audit

Every crawled page is checked, and the results roll up into one score from 0 to 100 across four pillars. How scoring works explains what moves it and what does not.

The important property: the score is over audited pages, not discovered ones. Pages nothing has fetched are missing from the measurement rather than counted as zero.

5. Publish

You write templates per category — a title pattern, a description pattern, structured data — and publish them. Generating content is not the same as publishing it; nothing is served until you publish.

6. Deliver

Your integration asks us for a URL's SEO and injects it into the response. The Deploy view reports what your integration saw, URL by URL: Active means traffic in the selected window with no errors, Idle means nobody requested the page, and Error means the integration reported a failure.

Idle is a statement about visitors, not about delivery. A page nobody asked for cannot report being served.

Where the loop closes

The next crawl sees what you published, the audit re-scores it, and the work log records the change. That is why the score line only moves when you crawl — every point on it is a crawl, not a day.