Your WordPress Site Looks Fine

What search and AI may be reading

A polished WordPress page can still publish an outdated description, an unrelated heading or structured data about the wrong service. Some problems are visible only when you inspect what the site actually sends to a browser.

During a recent WordPress and Elementor review, we found examples of all three, alongside broken links and nearly empty pages. We also found reasons to question the scanner’s interpretation. Here are the checks we recommend before making SEO or AI-visibility decisions from an automated report.

These observations establish problems in page output. They do not establish a ranking penalty, lost traffic or a missed AI citation.

Check hidden content before acting on a warning

One page contained an unrelated heading inside an Elementor section hidden at every configured screen size. It remained in the page’s underlying structure, while the browser’s visible content excluded it. A scanner included that hidden text in its assessment.

The text was worth investigating. Calling it the page’s visible opening was inaccurate.

Compare three views: the page a visitor sees, the HTML delivered by the server, and the rendered page structure after scripts run. Check desktop and mobile. In Elementor, review responsive visibility settings and hidden parent containers; its troubleshooting guidance explains where to look.

Preserve intentional mobile layouts and working accordions. Remove obsolete content only after confirming its purpose. This evidence does not tell us whether Google or a particular AI system used the hidden text.

Give metadata a clear owner

We found custom child-theme code outputting descriptions and social-sharing tags alongside an active SEO plugin. Editing the plugin field alone left another description in the published page.

Ask your developer to inspect every description tag, canonical tag and core social tag in the output, then trace each to its source. Decide which system owns each field. Preserve useful, accurate descriptions and page-specific images before disabling old output.

One approved description per page is a sensible cleanup target. Google may still choose its search snippet from the page’s text rather than that description, as its snippet documentation explains. Duplicate tags alone do not prove a ranking penalty.

Read what the schema actually says

Structured data is machine-readable information describing a page. Valid formatting does not guarantee correct meaning.

In this review, some accordion labels were marked up as FAQ questions even though they were service names or biography headings. Another page emitted questions about a different service.

Compare each question, answer and business detail with the relevant visible content. Google’s structured-data guidelines require accurate, relevant information and warn that automated tests cannot establish every quality requirement.

Remove inappropriate FAQ output without deleting useful page content. Do not generate filler questions simply to satisfy a scanner. Google retired FAQ rich results in May 2026, so this cleanup should not be presented as restoring that feature.

Follow the links and inspect the destination

A service card can look finished while its button leads to a missing page or an unrelated gallery. We found both kinds of problem. We also found category pages containing little beyond a heading and the shared navigation and footer.

Click the important service and contact links yourself. Confirm the final destination and have your developer check its HTTP status. A page displaying a “not found” message does not, by itself, prove the server returned a proper 404 response.

A category page should help someone choose their next step. Add a useful introduction and links to relevant, existing pages, or decide deliberately whether to consolidate or retire it. A page marked indexable is not necessarily indexed.

Repeated booking blocks also deserve a closer look. Verify which form or link visitors are meant to use, and whether the repetition serves a purpose. This is a usability check. Any search or AI impact remains unproven.

Figure 1. Two booking blocks appeared consecutively on the rendered page. Check whether both are needed before removing either. Actual browser crop, October 1, 2026.

Catch the leftovers from earlier edits

We found headings naming the wrong service, introductions describing a different offering, and stray [raw] and [/raw] tokens on public pages.

Figure 2. A literal shortcode appears as visitor-facing text. This actual browser detail was tightly cropped to exclude the surrounding content. Captured October 1, 2026.

WordPress documents how shortcode text can remain when no handler is registered, including after a plugin is disabled. That is a possible cause to investigate, not a diagnosis to assume.

Read the title, main heading, introduction and section headings together. Do they describe the same thing? Then check for leftover tokens and copied location names. Preserve images, captions and disclaimers during cleanup. Have the business verify prices, availability and consequential claims before rewriting them.

These are implementation and maintenance issues. The findings do not establish that Elementor itself is unsuitable for SEO.

Check what the scan actually covered

Crawler discovery, analysis and scoring may cover different URL sets. A WordPress inventory may also include internal builder templates rather than separate public pages. Several template records pointing to the homepage do not establish competing homepages.

Our automated review surfaced useful issues, while some findings needed manual correction or clearer explanation. Before comparing scores, ask:

• Which public URLs were analyzed successfully?

• Which were skipped, blocked, redirected or left outside the scan budget?

• Did the tool evaluate rendered content, raw HTML or stored builder data?

• Are both scans comparing the same URLs with the same rules?

Without those answers, a whole-site before-and-after score can be misleading.

Put the fixes in a useful order

Fix broken customer journeys and clearly wrong information first. Then resolve conflicting metadata and schema, improve empty destinations, and remove leftover content. Save a baseline and a rollback option.

After deployment, clear relevant caches and recheck the public pages. A local code change or saved editor field does not prove the live output changed. Test important forms without submitting inquiries.

An accurate, usable website supports SEO and AI discovery. Google says foundational SEO still applies to its generative AI features; no special AI schema is required. Bing recommends clear structure, supporting evidence and current business information in its AI search guidance.

Check crawler access separately. OpenAI’s documentation distinguishes OAI-SearchBot for ChatGPT search from GPTBot for potential model training. A green plugin score does not verify access.

Track search performance, reported citations and useful visitor actions over time. Establish that the right information is live before judging whether the changes improved visibility.

For related checks, see our AI recommendation diagnostic and SEO and AI visibility services.