You install a plugin, run a scan and get a reassuring score. What have you learned about your WordPress site?
You have learned something about its configuration. To find out whether people discover your content through AI search, you need different evidence. citelayer brings those local checks and observations into WordPress, so you can investigate a problem before changing content or buying another tool.
The useful starting point is to separate four questions: can a system access your content, has a matching crawler request reached WordPress, has a person clicked through, and does an external answer actually mention or cite you?
Start with a page you care about
Choose one public article or product page. Use a URL that visitors can open without signing in. After installing the free citelayer plugin, open its WordPress admin screen and run the readiness scan.
In Free v1.13.5, the site scanner checks six basic areas, with four additional checks when WooCommerce is active. Those include structured data, robots rules and endpoint health, alongside local content-structure and identity checks. Read each finding rather than treating the overall score as a prediction.
If an endpoint is unavailable, check its response. If a robots rule is restricting a crawler you want to allow, investigate that rule. If structured data is missing or incomplete, check the output from your existing SEO plugin before adding more markup. A site’s public response matters more than a ticked settings box.
Then test the content formats you have enabled. For a supported, in-scope page, citelayer can provide Markdown through an explicit .md URL, a ?format=markdown request or an Accept: text/markdown header. Open the requested output and check that it represents the page accurately. Private or intentionally excluded content should remain outside your published scope.
The quick start and Markdown configuration guide explain the relevant settings.
Understand what the score describes
The dashboard’s readiness score is a local diagnostic. It does not measure how frequently ChatGPT, Gemini or another service cites your site. The editor’s per-page checklist has its own method, so its number is not interchangeable with the dashboard score.
In the v1.13.5 dashboard scanner, llms.txt endpoint health is reported but does not contribute to the overall score. That distinction matters: a working optional output can be worth checking without being evidence of greater search visibility.
Google’s official guidance for generative AI search says that its search experiences continue to rely on Search’s ranking and quality systems. Google does not use llms.txt to improve Search visibility, and no special AI schema or Markdown file is required. Clear, useful public content and sound technical foundations remain the work to prioritize.
Read crawler logs as observations
Next, look at citelayer’s Bot Analytics and Crawler Logs. They record requests matching known AI-related User-Agent patterns when those requests reach the plugin’s collection path.
A logged match tells you that a request was observed. It does not prove that a provider indexed the page, used it in an answer or sent a customer. User-Agent labels are also not a verified identity check.
When a page appears in AI Blind Spots, read the finding as “no matching visit logged in this period.” A full-page cache or CDN may serve content without running WordPress, and retention limits the available history. Review cache compatibility before concluding that an unlogged page is inaccessible.
Look separately for people arriving
The AI Referrals view reports human click-through visits with recognized referral or UTM signals. Its totals answer a different question from crawler activity.
Open the local referral diagnostics when the numbers seem inconsistent. Check that tracking is enabled and that the browser can deliver the request. For a genuine referral visit, the Network panel can show whether delivery succeeded. Do not manufacture visits and then treat them as customer traffic.
Some real visits arrive without a recognizable AI signal. A copied link or a service that strips referrers can leave no attribution for citelayer to record. An empty referral report therefore needs context.
Pick the next action from the evidence
If a public endpoint fails, fix access first. If output is correct but an important page is unclear, improve its explanation and supporting evidence. If visitors arrive but do not understand what to do, work on the page’s next step. If you want to measure inclusion in external answers, use evidence from those systems and keep it separate from the local readiness score.
Start with the free plugin and quick start. Check one meaningful page, record what you find and repeat the same checks after your changes. That gives you a useful baseline for the work ahead.
