Findings & Fixes
Most tools end at "your schema markup could be improved". We end at the schema markup, generated, valid, and ready to paste.
What a finding is
One failed check, on one page or entity. Each carries a fix, a point value and a status. Not a category, not a severity band, a specific thing that is wrong in a specific place.
Product screenshotLocated
The URL, and the actual markup or text that failed the check, quoted as evidence. You should never have to hunt for what we meant.
Weighted
A point value derived from the rubric, so the list can be ordered by what actually moves the number rather than by how alarming it sounds.
Prioritised
Ranked by remediation priority, impact against effort, so the top of the list is where to start on Monday morning.
Stateful
Findings persist across crawls. You can see what was fixed, what regressed, and what has been sitting there for three months.
A finding only changes state on evidence. If a page was skipped, failed to fetch or wasn't sampled this crawl, its findings carry forward unchanged. Nothing is auto-resolved because we stopped looking, a "fixed" count that quietly includes pages nobody checked is worse than no count.
Artefacts, the fix generated
Findings attach generated output you can use directly.
Product screenshot| Artefact | What it is |
|---|---|
| JSON-LD | Structured data for the page, generated from its actual content |
| Answer block | A rewritten opening passage that leads with the answer, carries a statistic and can survive being quoted out of context |
| llms.txt | A valid file for your site, correctly sectioned, with descriptions that carry facts |
| robots.txt | Corrected directives where AI crawlers are being blocked unintentionally |
| Schema snippet | Targeted markup for a specific failing check |
| Content brief | Where the fix is editorial rather than technical, what the page needs to say, and why |
Asserted is not measured
Some inputs genuinely cannot be tested by crawling. Whether you publish original first-party research, for instance, or whether a quoted expert is real.
Those are recorded as declared, you assert them, and rendered with an explicit "asserted, not measured" badge. They are never quietly blended into a measured score, because a number that mixes the two is not a measurement of anything.
Export and integrate
Findings export to CSV with their evidence attached, so they can go into a ticket tracker or a client report without retyping. Visibility and Program plans include API access for findings, pages and rollups.
Related
Get fixes, not advice
Run a free audit and see the findings, and the artefacts, for your own site.