Score drivers

Why this matters

Every dimension score on an audit is the sum of many small contributions from individual findings. The drivers view shows you exactly which finding moved which dimension by how much — and what would move it back.

Where to find drivers

Open any audit. Below the dimension score row, click Show drivers. The panel lists, per dimension, the top 5 positive and top 5 negative contributors with the point delta each contributed.

How drivers are computed

For each finding, we calculate a point impact based on:

1. Severity. Critical / high / medium / low — a 0.5x to 4x multiplier.
2. Coverage. How many pages on your site the finding applies to.
3. Engine sensitivity. How much this kind of finding actually shifts AI engine behaviour, derived from our historical eval set.
4. Dimension weight. The base weight of the affected dimension in the headline score.

The product of these four numbers is the dimension point delta. Sum the deltas inside a dimension and you get the dimension's contribution to the headline.

Finding categories and which dimension they touch

Category	Primary dimension	Secondary
Missing sitemap	Discoverability	—
Broken internal link	Discoverability	—
Missing or wrong schema	Structure	—
Heading hierarchy issue	Structure	Clarity
Thin content (<200 words)	Clarity	Coverage
Vague claim without citation	Clarity	Authority
External link from unrelated site	Authority	—
Brand mention in news	Authority	Prompt presence
Lost rank on tracked prompt	Prompt presence	—
New gap — uncovered buyer prompt	Coverage	—
Comparison page missing	Coverage	Prompt presence

A finding can affect a secondary dimension at half weight when the link is causal but indirect.

Reading the drivers panel

Each driver row shows:

• The finding title.
• The dimension(s) affected.
• The point impact (e.g. +2.1 or -3.8).
• The page(s) it affects (linked).
• A Generate fix button when one applies.

What "Generate fix" does

It seeds a content brief or a publish task pre-configured to address the finding. For schema issues, that may be a publish task to your CMS connector with the schema block templated; for content gaps, it's a new draft in the right content type.

Why drivers can disagree with intuition

The dimension that "feels" worst isn't always the one losing you the most points. The drivers view is where this becomes obvious — a single critical-severity finding can outweigh a dozen low-severity ones. Trust the math.

Drivers vs. recommendations

Recommendations are the prioritised TODO list. Drivers are the explanation. The two views read off the same underlying findings — recommendations rank them by expected lift, drivers rank them by realised current impact.

Exporting drivers

CSV export available from the panel's row menu. Useful for executive reporting where stakeholders want to see "we earned +14 points on Structure this quarter by fixing X, Y, Z."