How Google understands a website: from crawling to the Knowledge Graph
By Anis Harbaoui, SEO specialist
Google understands a website in stages: it discovers URLs, crawls and renders them, indexes their content, extracts entities and relationships with natural-language processing, and connects those entities to its Knowledge Graph. Each stage can fail independently, which is why a page can be indexed yet misunderstood.
- Discovery depends on links and sitemaps; understanding depends on clear text and structure.
- Google can rank individual passages, so every section should answer one question clearly.
- Entities extracted from your text are reconciled with the Knowledge Graph; structured data helps that step.
The pipeline at a glance
| Stage | What happens | What you control |
|---|---|---|
| Discovery | Google finds URLs through links and sitemaps | Internal links, XML sitemap |
| Crawling | Googlebot fetches the page | robots.txt, server speed, status codes |
| Rendering | JavaScript is executed in an evergreen Chromium | Server-rendered content, no blocked resources |
| Indexing | Content is analysed and stored; duplicates are clustered | Canonicals, noindex, unique content |
| Understanding | Entities, topics and passages are extracted | Clear headings, direct answers, structured data |
| Ranking and serving | Results are chosen for each query | Relevance, quality, experience |
Discovery and crawling
Most URLs are discovered through links. A page that nothing links to (an orphaned page) may be found through the sitemap but receives no internal context. Keep important pages within three clicks of the homepage and list them in the XML sitemap.
Rendering and indexing
Google renders JavaScript, but rendering can be delayed and many other crawlers, including most AI crawlers, do not run JavaScript at all. Server-rendered HTML with the main text and JSON-LD in the initial response is the safest choice. During indexing, near-duplicate URLs are clustered and one canonical is chosen.
Understanding: entities, topics and passages
Google’s language systems identify the entities a page mentions, which one is central (its salience) and how they relate. Since 2021 Google can also rank individual passages of a long page for specific queries. Practical consequences:
- Put the main entity in the title, first paragraph and most headings.
- Start each section with a self-contained answer that makes sense out of context.
- Use tables for comparisons and ordered lists for procedures.
- Use
about,mentionsandsameAsto confirm what the text implies (see entity SEO explained).
The Knowledge Graph
Entities extracted from pages are reconciled with Google’s Knowledge Graph, a database of entities and facts. When your brand is reconciled, Google can attribute content to it and show a panel. Consistent identity data, Organization schema and independent coverage make reconciliation possible; see Organization schema explained.
Further reading: How Google Search works.
Frequently asked questions
Does Google read structured data before text?
No. It processes both. Structured data confirms and disambiguates what the visible text says.
What is passage ranking?
Google’s ability to rank a specific section of a page for a query, even when the page as a whole covers a broader topic.
Why is my page indexed but not ranking?
Indexing only means the page is stored. Ranking depends on relevance to the query, quality and competition.
Related guides
What is semantic SEO?
Optimise for meaning.
Entity SEO explained
Join the Knowledge Graph.
AI Overviews and WordPress
How AI answers use your pages.
NexalSEO adds the structured data and the sitemap that help Google read how your pages relate, and it is free.
See all features