Key takeaways
A semantic audit checks that each page of a site answers a real search intent. The page must use the expected vocabulary and entities, without competing with another page on the site. The audit starts from Search Console queries. It ends with a decision for each page: keep, optimise, merge, create or delete.
- One intent equals one target page: this is the rule that guides the whole audit.
- Cannibalisation shows up in Search Console, query by query.
- Semantic scoring tools help to sort; the decision remains an editorial one.
What is a semantic SEO audit?
A semantic audit analyses how a site’s content covers its subject. Which queries does each page target? Which intent does it answer, with what vocabulary? How do the pages fit together? It asks a simple question: does the site answer what your audience is really looking for?
It is the second area of a full SEO audit. It assumes a sound technical base. A page that is not indexed is not fixed by better text.
Semantic audit, keyword research, content audit: what are the differences?
| Approach | Question addressed | Starting point |
|---|---|---|
| Keyword research | Which queries are we targeting? | The market: volumes, suggestions, competitors |
| Semantic audit | Do existing pages answer the right queries, without duplicates? | The site: pages, Search Console queries, internal linking |
| Content audit | Are the texts readable, accurate, up to date? | Editorial quality: tone, sources, freshness |
In practice, the three overlap. Keyword research feeds the semantic audit, which then says which pages to create, merge or rewrite.
What is a semantic audit for?
- Align each page with one intent: informational, comparative, transactional or navigational.
- Remove cannibalisation: two pages competing for the same query hold each other back.
- Fill the gaps: topics your audience expects that are missing from the site.
- Structure the site architecture and internal linking: group pages by topic, link each supporting page to its pillar page.
- Make pages easier to cite: direct answers, explicit entities, dated sources.
Google does not ask for keywords. It asks for helpful, reliable content made for people (Google guide on helpful content, updated on December 10, 2025). The semantic audit checks that the site’s structure makes it possible to keep this promise.
How do you carry out a semantic audit in 6 steps?
1. List the pages and their queries
From Search Console, export the page × query pairs from the Performance report. Keep impressions, clicks, click-through rate and average position. Then isolate brand queries with the native filter launched in November 2025 (branded queries filter announcement). Semantic work focuses on non-brand queries.
2. Build the query × target page table
Group the queries by intent, then assign a single target page to each group. This table is the central deliverable of the audit.
| Query group | Intent | Target page | Competing pages | Decision |
|---|---|---|---|---|
| “seo audit”, “full seo audit” | Understand and launch an audit | /en/seo/audit/ | None | Keep, optimise |
| “technical audit checklist” | Check point by point | /en/seo/technical-seo-audit/ | Old checklist page | Merge, 301 |
| “semantic audit” | Content audit method | /en/seo/semantic-seo-audit/ | None | Keep, expand |
Qualify each intent from the actual SERP, not from intuition. Look at the types of pages at the top, the questions asked, the presence of an AI answer. Our guide to search intent explains the method in detail.
3. Measure topic coverage
For each target page, compare what it covers with what the query calls for. Rémi Mondina’s five-list method (Réacteur, April 2025) gives this work a structure. It lists entities, secondary keywords, variations, synonyms and contextual terms.
Google’s Cloud Natural Language API extracts the entities from a text. It gives them a salience score (entity analysis documentation). Version 2 of the API, released in August 2023, no longer documents this score (release notes). Stay on v1 and always review the output: the tool gets highly specialised terms wrong.
4. Detect cannibalisation and duplicates
Filter Search Console on one query and see how many pages get impressions. Two URLs that alternate in position on the same query signal cannibalisation. Complete this with the crawl: identical or very similar titles and H1s give away the same problem.
Keep the page that is older, has the links and matches the intent best. Move the useful content into it, then 301 redirect the other one. Google details the consolidation signals in its page on duplicate URLs.
5. Check tags, structure and internal linking
On each target page, check the title, the H1, the subheadings and the anchors of the links leading to it. Google builds its title links mainly from the title tag. It can also draw on subheadings or anchors (title link documentation). Consistent anchors and topic-based internal linking strengthen the signal. This is the principle behind semantic cocoons.
6. Write the editorial action plan
| Finding | Decision | Priority |
|---|---|---|
| Page in position 4 to 20 with many impressions | Optimise: tags, direct answer, internal linking | High |
| Many impressions, low click-through rate | Rewrite the title and meta description | High |
| Two pages on the same intent | Merge and 301 redirect | High |
| Page indexed with no impressions for months | Retarget, merge or delete | Medium |
| Searched intent with no target page | Create the page | Depending on potential |
Would you rather hand over this analysis? The audit is part of Elev8 Lab’s SEO and GEO services.
Which tools should you use for a semantic audit?
| Use | Tools |
|---|---|
| Real queries and pages | Google Search Console |
| Volumes and competitors | Semrush, Ahrefs, Haloscan, Keyword Planner |
| Structure and duplicates | Screaming Frog (title, H1, near duplicates, n-grams) |
| Entities and lexical field | Cloud Natural Language API v1, YourTextGuru, 1.fr |
| Users’ questions | Google’s People Also Ask, AnswerThePublic |
Warning
The optimisation scores of semantic tools are proprietary indicators, not Google criteria. A text can aim for 100 and still answer the intent poorly.
Which mistakes should you avoid during the audit?
- Reducing the audit to a list of keywords, with no decision per page.
- Aiming for a keyword density: the benchmark of about 1% is an agency convention, not a Google rule.
- Copying competitors’ structure: you end up with a paraphrase of the top 10, with no angle of your own.
- Creating one page per query variant: synonyms and plurals belong to the same intent.
- Forgetting internal linking: a well-written but isolated page stays weak.
Producing pages in bulk to cover every variant also exposes you to Google’s “scaled content abuse” policy. It targets content generated at scale with no added value (spam policies, updated on August 28, 2026).
Does a semantic audit matter for AI engines?
Yes. Answers in AI Overviews or from assistants reuse short, clear, sourced passages. Give each intent its own page, with a direct answer at the top and explicitly named entities. Your passages become easier to cite.
The semantic audit prepares this work. Techniques specific to answer engines fall under GEO optimisation. Once the content is aligned, popularity remains: that is the subject of the backlink audit.
Frequently asked questions
What is semantic analysis in SEO?
It is the study of the meaning of a text or a set of pages. It looks at the topics covered, the entities cited, the vocabulary used and the intent served. The semantic audit applies this analysis to a whole site to decide which pages to keep, optimise, merge or create.
What is a semantic field, with an example?
It is the set of words and notions associated with a topic. For “technical SEO audit”, you would expect for example crawl, indexing, robots.txt, sitemap, canonical, Core Web Vitals or logs. A page that leaves out these notions covers the topic poorly, even if it repeats the keyword.
What is the difference between a semantic audit and a technical audit?
The technical audit checks that Google can crawl and index the pages. The semantic audit checks that these pages answer the right queries, without competing with each other. The first comes before the second.
How long does a semantic audit take?
It depends mainly on the number of pages and queries to map. On a site of a few dozen pages, the inventory and the query × page table take a few days. On a catalogue of thousands of URLs, the work is done by template and by topic.
What are the 3 types of SEO?
There are generally three levers: technical (crawling and indexing), content (answering queries) and popularity (links). The semantic audit covers the second. The other two have their own audit: technical and backlinks.
Sources
- Google Search Central, Creating helpful, reliable, people-first content, updated on December 10, 2025. Accessed on September 24, 2026.
- Google Search Central Blog, Introducing the branded queries filter in Search Console, November 2025. Accessed on September 24, 2026.
- Google Cloud, Analyzing entities (Cloud Natural Language). Accessed on September 24, 2026.
- Google Cloud, Cloud Natural Language release notes. Accessed on September 24, 2026.
- Google Search Central, How to specify a canonical URL with rel=”canonical” and other methods. Accessed on September 24, 2026.
- Google Search Central, Influencing title links in Google Search. Accessed on September 24, 2026.
- Google Search Central, Spam policies for Google web search, updated on August 28, 2026. Accessed on September 24, 2026.
Cite this article
, . (2026, September 26). Semantic SEO audit: method and tools. Elev8 Lab. https://elev8-lab.fr/en/seo/semantic-seo-audit/