ContentGrapher
AboutResearchHelp Docs
Analyze →
ContentGrapher
Analyze →
AboutResearchHelp Docs
← All reference pages

Coverage Score

How the score is computed from integration quality and depth-weighted question coverage, and what it does not measure.

What the score measures

The coverage score is a composite measure of how well your content is structured for the topic ContentGrapher detected. It has two components.

The first component (half of the score) measures concept integration quality: the ratio of your observed concepts that are fully integrated — well explained, consistently named, and connected to related ideas — to the total number of concepts ContentGrapher identified in your content.

The second component (the other half) measures diagnostic question coverage. ContentGrapher asks 8 structural questions about your primary topic: what the topic is, how it works, what it depends on, what it produces, who interacts with it, what its constraints are, what alternatives exist, and what a grounding example looks like. Each question is rated on a four-level depth scale: not addressed, mentioned in passing, explained with a mechanism, or grounded with a worked example or named entity. Each rating is then refined by signals in the relevant passage of your content: whether the answer has enough length to develop, whether it includes a worked example, and whether it names concrete entities. The 8 refined ratings combine to produce this component, giving the score finer resolution than a simple count of covered questions.

The two components count equally. If question depth could not be scored for an analysis — for example, an older analysis whose source page can no longer be re-fetched — the score reads "Integration only · depth not scored" and reflects concept integration alone. Re-analyzing the page adds depth scoring and restores the full blend.

A score of 74% means your content's combination of integration quality and depth-weighted question coverage reaches 74% of the maximum on these two dimensions.

In the report, the score shows at the top, beside the page title, under "Measured coverage". Below it, the report shows how many concepts are well integrated (for example, "7/9 well integrated") and the anchor-question depth as a percentage. The note "Score only · not a list of edits" means the score measures the page as it is; the next steps are a separate list.

The score block at the top of a report: a ring with 17%, then "Measured coverage", "0/3 well integrated", "33% anchor-question depth" and "Score only · not a list of edits".

Why does my report say "Coverage score not available"?

The report shows "Coverage score not available" in place of the percentage when it cannot compute the score. This happens when the report holds no question ratings for the page, or when the analysis found no concepts on the page.

The rest of the report still shows what the analysis found: the concept map, the page overview and any next steps. If the report holds no question ratings, re-analyzing the page produces them.

Reading the percentage and the Done badge

The report shows the score as a percentage. It does not place the score in a named band. Use the percentage to compare analyses of the same page over time, not as a target.

In your tracked pages on the dashboard, ContentGrapher marks a page done when three conditions hold: the coverage score is 70% or higher; every concept classified as essential in the explanation framework is fully covered, not partial or missing; and the page-structure advice is not to split the page into multiple pages. A "Done" badge then appears next to the page.

Once a page is marked done, a small dip on a later re-analysis does not immediately undo that status. Done only reverts on a clear regression: the score falling well below 70%, or an essential concept becoming missing again. This keeps the done status stable against the kind of small run-to-run noise described in the delta view, rather than flickering on and off between analyses.

The explanation framework is a separate measure

In Phase 2, ContentGrapher also builds an explanation framework for your topic and audience: the complete set of concepts that should exist on a page like this, with each concept classified as essential, important, or useful. The next steps come from the gap between what you have and what this framework includes.

The report shows this gap in two places: on the concept map, where an outlined concept is a proposed addition (the page overview lists them under "Concepts to look at", with a dashed outline); and in the next steps. It is a different measure from the coverage score.

The coverage score reflects the structural quality of what is already in your content. The framework gap reflects what is missing compared to an ideal page for your specific topic and audience. The score tells you where you are. The next steps tell you what to do next.

How integration state and question coverage work

A concept is well integrated when it is named consistently, explained at sufficient depth, and connected to related concepts explicitly. The report labels it "Well integrated". Three other states indicate partial or incomplete coverage: "Lightly explained" (present but not fully explained), "Needs more explanation" (named but lacks depth), and "Naming varies" (the concept appears under different terms across the page). On the concept map, a filled dot marks a well-integrated concept and an amber diamond marks a concept that needs attention.

Only well-integrated concepts contribute to the integration half of the score. The question coverage half is independent: it measures how thoroughly your content addresses the 8 diagnostic questions about your primary topic, with each question rated for depth on a four-level scale, regardless of how many total concepts you have.

What the score does not measure

The coverage score is not an SEO score. It does not measure keyword frequency, readability, word count, or competitor alignment.

It does not measure whether your content is accurate, engaging, or well-written. You can reach a high score with technically correct but poorly written prose.

It does not benchmark against other pages. ContentGrapher derives its analysis from your topic and audience, not from what currently ranks.

A score of 100% means structural completeness as ContentGrapher's model defines it for this topic and audience. It does not guarantee citation — domain authority, publication, and indexing are outside the model.

How to use the score

Use it directionally, not as a target. The goal is not 100% — it is a measurable improvement in the gaps that matter most for your audience.

Track the score across re-analyses using the delta view. A score change alone does not establish that your revisions addressed the gaps: repeated analyses can differ even when the source is unchanged. Check the comparison against your edits and the next steps.

The guidance to start with flagged gaps is grounded in the Decoy Study, ContentGrapher's published controlled test across 40 third-party pages. In that study, on pages with five or more flagged gaps, filling the flagged gaps improved AI retrieval by 11 percentage points more than adding the same amount of content to parts the analysis said were already fine. Those figures are specific to the study's pages and retrieval setup; the size of the effect on your page will differ.

The gap count itself is part of the signal. The Decoy Study found that when a page has five or more flagged gaps, addressing the specific gaps identified made a measurable difference: targeted additions outperformed untargeted structural additions by 11 percentage points on the study's retrieval measure. When a page has two or fewer flagged gaps, the study found no meaningful difference between addressing flagged gaps and adding content to areas the analysis said were already fine. Those figures are specific to that study's pages and retrieval setup; the effect on your page will differ.

Related topics

→Delta View→Next steps→How do I read the concept map?
← All reference pages