An SEO plugin handles the obvious. The problems below came from three WordPress sites, one of them with an SEO plugin installed.
On this page
AI SEO for WordPress is mostly about the things an SEO plugin cannot see: whether AI crawlers get past the host, whether the structured data is telling the truth, and what the theme and page builder are shipping.
Three of my eight case studies are WordPress sites: a WooCommerce food store, an Elementor site for a B2B software company, and a B2B telecom site on a page builder with an SEO plugin.
The telecom site had an SEO plugin installed. Its structured data still declared the business as both a person and an organisation, with no logo or contact details, and a self-serving star rating was marked up on the homepage, which is a review-snippet policy risk.
On the WooCommerce store, most of the URLs submitted in the sitemap were near-empty tag archives, indexable and crowding out the real catalogue.
WordPress has one real advantage for AI search: it renders pages on the server, so the content is in the HTML. The failures happen around it. On the Elementor site, robots.txt invited AI crawlers and the host's firewall blocked them, and every image was lazy-loaded by JavaScript.
Your developer, or AI under my direction. Each fix in the plan comes with a way to prove it landed. On the WooCommerce store, the client's own team made them.
Real engagements, described by sector only. Each includes the report and the workbook.
Full-site technical audit, then an organic growth plan and a topical-authority cluster architecture, with a forecast model built on leading indicators rather than invented targets.
Three automated warnings were rejected before publishing — including a 'duplicate title on every page' that turned out to be the titles inside SVG icons.
Audit delivered; no live-site changes recorded.
Full technical, content, schema and AI-search audit, then two re-audits to track what the client's team fixed, followed by keyword research and a competitor gap analysis.
Four first-draft claims were withdrawn or corrected before the report was released — including two 'duplicate product' pairs that had only been inferred from similar-looking URLs.
Audit delivered, then re-audited twice.
Full technical, schema, performance and AI-search audit with a phased action plan. When the top finding was challenged, I re-tested it against independent third-party crawl data — it held.
Two candidate findings were dropped in verification — one was a redirect the checker had followed and mistaken for a live duplicate page.
Audit delivered; fixes not yet implemented.
Mine runs in two passes. The first is an agentic audit of the whole site through systems I designed: indexation, canonicals, sitemaps, structured data, performance, images, AI crawler access, and the copy itself checked for factual errors.
The second pass is the one that matters. Every finding is re-tested by hand against the live site and labelled Confirmed, Likely or Unverified before it goes in the report, with a check you can run yourself.
Yes, and I do: the first pass of every audit runs through AI-directed systems. Use it for breadth, never for the final word. Anything an AI tool reports needs checking against the live site before anyone acts on it.
Send a URL. Three to five verified findings come back within two working days, free, with no call required.
Ask my AI
Answers from notes I wrote. It shows its sources and says when it does not know.
Ask about the case studies, how I would approach your site, or what actually goes into an audit. Every answer shows the note it came from.
AI-generated from my own notes. Verify anything that matters.