Semantic SEO Audit: How to Check Topics, Intents, Entities and Links

Since Google introduced its Knowledge Graph in 2012, announcing a search that understands "things, not strings" (more than 500 million objects and 3.5 billion facts at launch), ranking has depended less on repeating a keyword and more on covering a subject completely. A semantic SEO audit measures exactly that: how well your site, page by page and as a whole, covers its topics. This guide gives the 7-step method, the tools for each step, the decisions to take and the deliverable to produce.

What is a semantic SEO audit?

A semantic SEO audit is the analysis of the meaning of a site's content: the topics it covers, the search intent each page answers, the entities it names and the links that connect pages on the same subject. It applies the principles of semantic SEO to an existing site. It answers four questions:

1. Which page owns which query? Each important query should have one, and only one, page built to answer it.
2. Does each page answer its intent completely? Subtopics, questions and examples that the searcher expects.
3. Which topics and entities are missing? Compared with the searcher's needs and with the pages that rank.
4. Do the links reflect the topics? Pages on the same subject should link to each other and to a pillar page.

Semantic audit, technical audit, content audit: what is the difference?

The three audits overlap and feed each other. A content audit judges each page on its own; the semantic audit judges how pages relate to queries and to each other.

When should you run a semantic audit?

• Several pages rank in turn for the same query, or none of them reaches the first page.
• Traffic is flat despite regular publishing: new articles repeat old ones instead of covering new subtopics.
• Before a redesign or a migration, to decide which pages to merge and how to organize the new structure.
• Before a content plan, to write what is missing rather than what already exists.
• When AI assistants cite competitors on questions your site should answer.

How to run a semantic SEO audit in 7 steps

1. Inventory the pages and their queries

Crawl the site to list every indexable URL with its title, H1 and headings. Then export the queries per page from the Search Console Performance report (impressions, clicks, average position) over at least three months. Join the two lists in one spreadsheet: one row per URL, with its top queries.

2. Assign one primary query and one intent to each page

For each URL, write the primary query it should own and its intent (informational, commercial, transactional, navigational). Check the intent in the search results: if Google shows guides and your page is a product page, the page is misaligned. Our guide to the types of keywords details the intents. The rule: one intent per page, one page per intent.

3. Detect cannibalization and misalignment

Sort the export by query and look for queries where two or more of your URLs collect impressions. If the URLs alternate in the results over the weeks, you have keyword cannibalization: Google cannot tell which page to rank. Note also pages whose main queries do not match the topic they were written for.

4. Check topic coverage page by page

For each important page, compare its headings with the H2 and H3 of the top five results and with the People also ask questions. Subtopics that every ranking page covers and yours does not are gaps to fill. These subtopics are the page's secondary keywords. Google's guidance on helpful content asks the same question: does the content provide "a substantial, complete, or comprehensive description of the topic"?

5. Audit the entities

Entities are the people, products, places, brands and concepts a page talks about. List the entities a reader expects on the topic (for a page about running shoes: cushioning, drop, pronation, the main brands) and check that the page names and explains them. The entity analysis of Google's Natural Language API returns the entities detected in a text with a salience score, a quick way to see what a machine thinks your page is about. Our glossary explains entity SEO in detail.

6. Find the topic gaps at site level

Draw the topical map of your subject: the pillar topics and all their subtopics. Place each existing page on the map. The empty boxes are your content gaps, the pages to create. Group candidate keywords by search results overlap with our free keyword clustering tool to know whether a gap needs a new page or a section in an existing one.

7. Review the internal linking by topic

Pages on the same topic should link to each other and to their pillar page, with descriptive anchors. Check that each important page receives links from its siblings and that no page is orphaned. Our free internal link analyzer lists the links of a page, and our guide to the semantic cocoon explains how to organize links by topic.

The 7 steps at a glance

What to do with each page: the decision grid

The audit is only useful if every page ends with a decision. Five are enough:

Prioritize by impact and effort: merging two cannibalizing pages and enriching pages ranked between positions 5 and 20 usually move fastest, because Google already associates them with the query.

What should the deliverable contain?

• The query-to-page map: one row per page with its primary query, intent, secondary keywords, main entities, pillar page and decision.
• The conflict list: cannibalized queries with the URL to keep and the URL to merge.
• The gap list: pages to create, grouped by topic cluster, with the target query of each.
• The linking plan: links to add between pages of the same cluster, with the anchor.
• The backlog: all actions sorted by priority, each with a success criterion (for example, a single URL ranking for the query after two months).

This map is also the base of your topical authority: a site that covers a subject completely, without duplicates, is the one Google and AI assistants trust on that subject.

Semantic audit for SEO and GEO

AI assistants answer a question by splitting it into sub-questions and looking for passages that answer each one. Google describes this query fan-out technique in its documentation on AI features, and states that there are no additional requirements to appear in AI Overviews or AI Mode. The work of a semantic audit is therefore the same for both: one page per intent, one section per sub-question, entities named explicitly. A page that covers its topic in self-contained sections ranks on more queries and offers more passages an AI can quote. A site without cannibalization also sends a clearer signal of which page to cite.

Common mistakes in a semantic audit

1. Auditing keywords instead of intents. Two different keywords with the same intent belong on one page.
2. Creating pages for every variant. A variant is a section, not a page; our guide to keyword clustering shows how to decide.
3. Stopping at the diagnosis. Without a decision per page and a backlog, the audit stays a file.
4. Forgetting the links. Merging and creating pages without updating internal links leaves new pages orphaned.
5. Never repeating it. Each new article can create a new conflict; a quarterly check of Search Console catches them early.

From audit to action

A semantic audit turns a pile of pages into a map where every query has one owner and every gap has a plan. The hardest part is keeping the map alive as the site grows. Sorank finds the queries of your site automatically, its topical cluster generator organizes them into clusters, and its blog article generator writes and publishes the missing articles with internal links, in autopilot. Plans start at $49 per month on the annual plan, with a 3-day free trial.