Know what a page covers, and what it still lacks for its query
A page can be well written and still miss the query it aimed for. We put its actual content against the keyword you gave it, count on every one of your pages the signals engines expect from credible content, and group your keywords by meaning so you can see which page should carry what.
Access opens in waves: we email you when yours is ready.
Our numbers, with their provenance
-
5finding blocks per page report measured
-
6E-E-A-T signals counted per page measured
A well-written page can still aim beside the point
A text can be clear, sourced and pleasant to read, and answer a different question from the one that brings you traffic. The three situations below are fixed by rereading the page, provided you know which page and why.
-
01
You write for one query, the page serves another
Intent drifts while writing: a comparison becomes a guide, a service page becomes an article. The target keyword, meanwhile, is still the one from the brief.
-
02
Two of your pages aim at the same keyword
They compete for the same query, split the signals between them, and the engine picks the one you would not have picked. Nobody notices before comparing the pages one by one.
-
03
A brief says what to write, it stops at what is missing
After publishing, the question becomes “what does this page still fail to cover?”. Answering it means rereading the text against the query’s semantic field, which nobody does for a hundred pages.
What we measure
Two natures of finding, never blended
A page report is the structured opinion of a language model about your text. The E-E-A-T signals are counted by code, page by page, and each carries its denominator. Both serve the same decision, and they are shown separately because they are not established the same way.
- The page report against its keyword (detected intent against expected intent, the semantic field with occurrences found and recommended, title, H1, meta description and subheadings each with a proposed rewrite. The number of finding blocks is published in our claims registry.)
- E-E-A-T signals, counted page by page (identifiable author, structured data, cited sources, social sharing tags, encrypted connection, text volume. Every signal appears with the number of pages concerned and the total examined, never a bare percentage.)
- Your keywords grouped by meaning (queries that mean the same thing are brought together, then the grouping is put against results pages already captured: two queries serving largely the same domains aim at the same intent, on the engine’s own account. The match is counted, with its denominator, rather than quietly recut.)
- The candidate page, and the cannibalisation risk (among the pages actually crawled, the product proposes the one that best fits a keyword, and flags the others aiming at the same query. The ranking is deterministic: it replays identically.)
Further reading
Semantic coverage: what can be measured on a page, and what can only be judged
The guide: what can be measured on a page, and what stays an appreciation.
Read ComparisonNessFlow vs Semrush
What an all-in-one suite judges about content, and the data it judges from.
Read GlossaryGlossary
Semantic coverage, search intent, structured data: the definitions.
ReadOther modules
Frequently asked questions
Who produces the analysis of a page?
A language model, from the page’s actual content as captured by the audit: the title, the subheadings, the meta description and the text. Its output is structured, so it compares from one page to the next, but it remains an opinion. That is why the page says so instead of presenting it as a measurement, and why the counted signals are shown separately.
What is the difference between a page rating and the E-E-A-T signals?
The rating comes from the model: it situates a page against its keyword, and two analyses of the same page can move it by a few points. The E-E-A-T signals are counted by code on each of your pages, with the examined total next to them: two passes over the same audit return exactly the same number. The first helps you prioritise, the second can be quoted in a client report.
Does an analysis need a finished audit?
Yes, and that is deliberate. The analysis covers the content an engine actually obtained from your page, JavaScript rendering included, rather than text pasted into a form. An audit also provides the project’s other pages, which is what makes pages aiming at the same query visible: without it, cannibalisation would stay invisible.
How are keywords grouped by meaning?
By meaning proximity computed on vectors, with a threshold set on real pairs rather than picked by eye. The grouping is then put against the results pages of your own tracking: two queries whose results serve largely the same domains aim at the same intent. That check returns a count with its denominator, and it does not recut the groups: only tracked keywords have a reading, and the shape of your content plan must not depend on your tracking list.
The scope of this module, and how it is triggered
One analysis covers one page and one target keyword, it runs on request, and it needs a finished audit to read the page’s actual content. The scope below follows from that, and writing it beats letting anyone imagine an automatic proofreader.
What we cover
- One report per target keyword, run on request against a finished audit
- E-E-A-T signals counted on every crawled page, each with its denominator
- Your keywords grouped by meaning, and a count of what the results pages confirm
Stated limits
- A page rating comes from a model: it situates the page, it does not measure it
- No site rating: this module returns an opinion per page, which the scorecard leaves out of its calculation
- No automatic writing: titles and meta descriptions are proposed, the body of the page stays written by your team
- A number of analyses per month depending on the plan, and a run on request: nothing starts on its own
What each page covers, and what it lacks
One report per target keyword, signals counted with their total, and a grouping by meaning put against real results: enough to decide which page to rewrite.
Access opens in waves: we email you when yours is ready.