Skip to main content
LLM audit score cards and audit history
Use Technical → LLM audit to identify access and content problems that can make your site harder for AI systems to read. Open a completed history row to inspect its evidence. Screenshots use illustrative local audit data.

Read the headline scores

Performance, Accessibility, Best practices, and SEO show Lighthouse category results on a 0–100 scale. Content is a separate assessment of readability, heading structure, and entity coverage. A ten-point change in one category is not equivalent to ten points in another. Compare like-for-like runs and read the findings before choosing a fix. These diagnostics do not predict rankings, citations, or conversions. Complete an SEO audit before a new LLM audit when you need the site URL inventory and page-level evidence.

Overview tab

LLM audit Overview tab showing Website performance and crawlability evidence
Start with Website performance, then inspect crawlability and the recorded technical findings. Report sections depend on what the run captured; an older report can have less evidence.
  • LLM crawlability: inspect the captured robots.txt, sitemap, and llms.txt results and bot access rules. A sitemap lists discoverable URLs; it does not prove indexing. A missing optional llms.txt file is different from an access block.
  • Technical AI accessibility: review each check’s status, priority, details, and recommendation. Domain-level checks do not establish that every page was tested.
  • Content findings: use the content overview to identify recurring weaknesses and open the corresponding page evidence. A selected finding carries its filter into the Pages tab.
  • Source evidence: inspect recorded initial HTML and NLP details when available. Initial HTML can differ from the later browser-rendered page.
  • Summary: when supplied, use the action brief to prioritize work, then confirm each action against the recorded findings.
Fix confirmed access blocks first. Then assign specific content and technical recommendations to the relevant page owners. Use Crawler logs for observed bot requests; an audit is a snapshot, not a visit log.

Pages tab

LLM audit Pages tab with the site URL inventory and page-level findings
Use Site URL overview to inspect the captured page inventory. Compare HTTP, Fetch time, Words, and Content recommendations. Search and sort the list to find the affected URLs. Unknown values mean evidence is unavailable, not a zero score. Expand a page to review SEO findings, content recommendations, and recorded source details. Recommendations are ordered by priority. Expand a recommendation to see its explanation and the page values supporting it. Passages to improve, when present, shows the excerpts that need attention. When arriving from a content finding in Overview, review the matching pages and clear the filter to return to the broader inventory. Additional recorded pages contains source evidence outside the SEO URL inventory when available. If the report says page-level findings were not captured, complete an SEO audit and run a new LLM audit. Do not interpret a missing inventory as a clean site.

How these scores are calculated (simple)

Lighthouse supplies the Performance, Accessibility, Best practices, and SEO categories. Content combines readability (35%), heading structure (30%), and entity/keyword coverage (35%). The underlying checks explain what to change; scores alone do not establish whether an agent can complete a task. Use Agent Experience (AX) for recorded agent tasks and journey outcomes.

Turn findings into work

  1. Confirm the target URL and captured evidence before creating a ticket.
  2. Resolve access failures and high-priority page findings first.
  3. Assign a concrete page or technical change, with an owner.
  4. Rerun after the change and compare the relevant findings, not only the overall score.

Where to go next