A business can have strong reviews and useful services while its website leaves basic questions unanswered. Before proposing a redesign, we check two things: whether the relevant search system can access the content, and whether that content accurately explains the offer.
These checks address information gaps. They do not guarantee that ChatGPT, Gemini or Perplexity will recommend the company.
Check access for the system you are testing
Googlebot can render JavaScript. A page is not invisible to Google simply because its text loads through JavaScript. Blocked resources, rendering errors and content that requires a user interaction can still cause problems, so use the URL Inspection tool in Search Console to check what Google actually receives and renders.
Other systems have their own crawlers and access rules. Check the relevant documentation, robots directives and server responses rather than assuming every crawler behaves like Googlebot. A sitemap helps search engines discover URLs, but it does not guarantee indexing.
Check whether the facts are current
Open a priority service or product page and look for the details a customer needs before contacting you: scope, eligibility, location, pricing context and the next step.
Confirm that prices, opening hours and contact details are current. Where an answer is missing, add a factual answer that the business can confirm. Clear content is more useful to customers and easier to reuse in search answers, but which sources an AI assistant selects remains outside your control.
Treat structured data as a description
Where structured data is used, it should match the content visitors can see. Choose supported types that fit the page and verify the details, rather than adding every schema type to every page.
Google states that no special schema.org markup is required for AI Overviews or AI Mode. Structured data does not guarantee a rich result, a citation or a business recommendation. Google’s guidance applies to its own Search features and should not be presented as a description of every AI service.
What we observed in client work
For a local e-commerce client, we clarified product information, answered buyer questions and fixed the technical structure, while local SEO and Maps work ran alongside. By month three, inquiries were well above the starting point, and visitors referred by AI tools had started sending inquiries.
For a chiropractic clinic, we rewrote key pages around patients’ questions while local SEO work continued. By month three, inquiries had increased and some new patients mentioned AI tools.
These observations came from combined work. They do not isolate schema markup as the cause, and they do not show that a particular assistant began recommending either business.
Record the checks and the outcomes separately
When testing AI mentions, save the query, date, system and answer. Track identifiable referrals and qualified inquiries separately. A referral is not the same as a verified recommendation, and a single answer is not a stable visibility score.
Start with access, inaccurate facts and missing answers on priority pages. An IGM audit can turn those findings into a practical list of corrections. Request one at pro-igm.com.
Frequently asked questions
Does JavaScript make a page unreadable? Not automatically. Googlebot can render JavaScript. Verify which content is available to the specific crawler, and fix any access or rendering problems.
Will schema markup make an assistant recommend us? There is no such guarantee. Markup must describe real, visible content and meet the requirements of the relevant system.
How soon should we expect a recommendation? The cases above describe observations around month three, not a promised deadline. Keep tracking mentions and actual inquiries over time.