Have you added schema markup to your site expecting ChatGPT or Google’s AI Overviews to start citing your content? But unfortunately, nothing has worked? Before you wonder whether the markup is broken, let me reassure you it isn’t. You see, structured data does help with AI search, but less directly than you think.
At B3NET, we’ve been building and optimizing websites since 1999, and honestly the game has changed forever. We’ve spent long enough tracking meta keywords before they become worthless and FAQ dropdowns come and go from Google’s results. That kind of result makes us skeptical of any claim that one markup format decides who AI tools cite.
Let’s break it down.
The short answer
Structured data shows search engines and AI systems what your content is about and who stands behind it. Whether you get cited is another question that depends on a bunch of other factors, like page crawlability, whether it’s indexed, whether the engine trusts it, and whether it answers obvious user queries without beating around the bush. Schema helps a good page get read and found correctly, but it won’t rescue one with thin content.
What rich snippets taught us?
When Google, Bing, and Yahoo launched Schema.org in 2011, the idea was simple. Label your content so machines and bots don’t have to guess. That means a product price gets tagged as a price, a review score as a rating.
The reward was how your listing looked, with star ratings and recipe cards, not where it ranked. Google’s John Mueller has said plainly that structured data won’t make a site rank better across SERPs. Even the visual reward is temporary.
In 2023, Google restricted FAQ rich results to authoritative government and health sites and removed HowTo rich results completely. As a result, sites with hundreds of marked-up pages went down within a few weeks.
We took a leaf out of that, which is: a markup is only valuable when it describes what’s on the page. But a markup added to chase a search feature lasts as long as the feature does itself.
So, how does AI search find your content?
AI answer engines don’t just read your JSON-LD and paraphrase it back. ChatGPT, Perplexity, and Google’s AI Overviews pull pages from a search index and generate a unique answer. Roger Montti made this point in Search Engine Journal, and we know for sure that search engines do separate the main content from headers and footers.
On the other hand, Google’s guidance on AI Overviews shows you don’t need any special markup beyond normal SEO best practices. Microsoft has added to it with Bing’s Fabrice Canel confirming that Bing uses Schema.org markup to help its models, including Copilot, understand page content. In fact, ChatGPT’s searches have drawn heavily on Bing’s index.
The bottom line: Schema influences how a search engine classifies your page before any AI model retrieves it. While it’s an indirect factor, it’s still worth having.
Where schema helps most?
AI systems are trying to pin down details and facts, like who published a page, who wrote it, what the product costs, and whether it’s in stock. Structured data is most useful when it answers such questions in the very same manner as your site does.
For example, take a blog post from an ecommerce site, and you’ll see how most implementations tag it as an Article. But a stronger setup would link the article to the actual author, preferably someone with a profile page, or the company that publishes it. It’s easy. You define the organization once with an @id and reference it everywhere else:
“publisher”: { “@id”: “https://www.example.com/#organization” }
If the company later changes its name or logo, you can always update one definition, and every page will pick up the change. Helen Pollitt, Head of SEO at Getty Images, describes the same in Search Engine Journal, where he vouches for treating schema as an entity strategy instead of a checklist.
What you should do: As a business site, you should start your schema markup with Organization, Person for authors, Article, and Product with Offer. Add LocalBusiness if customers visit in-person.
4 mistakes that spoil AI visibility
Mistake 1: Markup that contradicts the page.
Schema claiming a 4.9 rating from 127 reviews for a product page with no visible reviews on screen is sure to confuse AI systems, and even call for a Google manual action.
Mistake 2: Working with stale values.
If a schema says $59.99 and the page says otherwise, like $79.99, an AI system faces conflicting facts from one URL and therefore has less reason to trust either.
Mistake 3: Identity drift
Product templates designed earlier can still carry an old brand name in their Organization markup. This is in sharp conflict with the homepage carrying a new markup.
Mistake 4: Assuming your site is the only source
If directories or old press coverage still lists outdated names or prices for your brand, correcting markup on your own pages may not help. Instead, clean up the listings all throughout.
Where to begin?
Not just theory, but actual steps you can take right now to make things better with structured data for AI mentions.
Step 1: Run your main page templates through Google’s Rich Results Test and the Schema.org validator.
Step 2: Check each marked-up value against what a visitor sees.
Step 3: Define your Organization and author entities with consistent @id values, and reference them across the site.
Step 4: Make sure robots.txt isn’t blocking the crawlers that feed AI search: OAI-SearchBot (ChatGPT search), PerplexityBot, and Bingbot (which also supports Copilot). GPTBot collects training data for OpenAI’s models, so blocking it is a separate decision that won’t remove you from ChatGPT search results.
Step 5: Use IndexNow to get updated pages to Bing faster.
Step 6: Spend time improving the content that answers your customers’ questions clearly.
Not sure what your markup is telling search engines? B3NET’s free SEO audit will show you. If you’re ready to fix it, request a proposal from our SEO and AEO team.
