Page titles
Missing titles and repeated title text are review flags. Length checks use 15 and 60 characters as rough heuristics, not search-engine limits or a prediction of truncation.
[ method ]Evidence before conclusions
Know what your audit can tell you.
An automated preview is a starting point for review. We show what was observed, identify what was modeled, and keep the limits visible.
Method: free-readiness-preview-v2 · Published October 25, 2026. Each report records the version used; older reports may use v1 or have no recorded version.
A bounded public-page review
The preview fetches up to five public pages, beginning with your submitted URL and following eligible same-site links. It reads returned HTML without executing JavaScript, signing in, or inspecting your analytics. Your report lists the pages actually fetched and the completion time when recorded. An unreachable page is not evidence that its content is missing.
What the checks observe
Headings and content
We count H1 headings separately from other headings. Fewer than 200 extracted text characters prompts a low-priority review of the rendered page. It does not prove poor content quality.
Descriptions and structured data
We check for a meta description, JSON-LD or microdata, and malformed JSON-LD. Syntax parsing does not validate schema semantics, factual accuracy, or eligibility for enhanced results.
Discovery files
A robots.txt absence flag requires an observed 404 or 410 response. Timeouts and access failures are unknown, not evidence of absence. An absent llms.txt is not a readiness defect.
Scenarios are hypotheses
A general language model generates up to five buyer questions from the fetched copy and one answer sample per question in the public preview. These are not live queries to ChatGPT, Claude, Google AI Overviews, or other named products. Brand absence in one sample is not a visibility score, citation rate, ranking, or market-share estimate. Model output can change between runs.
A label of “measured” on a technical recommendation means its source is a deterministic check of fetched HTML. The impact and recommended fix still need human review. Sample drafts and inferred writing style are generated suggestions: verify every business claim before publishing.
How we order next steps
- Group equivalent actions while retaining their distinct page evidence.
- Order technical observations by configured severity: high, medium, then low. These are triage rules, not measured financial impact.
- Keep modeled content hypotheses at low priority, after measured observations at the same priority.
- Use title order for stable ties and return up to ten actions. Review the full findings list for observations outside that shortlist.
There is no overall score or scoring weight in this automated preview. The separate self-assessment scorecard and paid human-reviewed audit use their own disclosed methods.
What remains outside this preview
This is not a full-site crawl, JavaScript rendering audit, accessibility certification, Core Web Vitals measurement, backlink analysis, or verified multi-engine visibility study. No findings does not mean the website is error-free. Reports expire within 24 hours; download a copy if you want to keep it.
Version 2 retains evidence across pages and treats short HTML, absent descriptions, and absent structured data as low-priority review items. Version 1 used stronger severity for those heuristics. A change in findings across versions may reflect methodology rather than a website change.
Reference guidance
Character counts are review heuristics. Google may choose different title or snippet text; a particular length does not guarantee how a result appears.