Schema evidence problem
The schema evidence problem is that nobody has shown structured data makes AI answers cite you more, even though it's often sold as if it does.
The schema evidence problem is the gap between what people say about structured data and what's been shown. Schema markup is often sold as a way to get cited more in AI answers. When people have tested that, the effect has been small or hard to find.
It matters because time spent on markup is time not spent on the things that do move answers, like what your pages say and who else writes about you.
What the evidence says
Google says there's no special schema.org markup you need to add to appear in AI Overviews or AI Mode, in its guide to AI features. It also says any markup you have should match the text people can see on the page.
Ahrefs tracked pages that added schema and compared them with similar pages that didn't. It found no real lift in citations on any platform it checked. It also pointed out why the opposite belief took hold: sites that use schema tend to be well built in other ways too, so they get cited for those reasons.
So should you bother?
Yes, for the right reasons. Structured data helps search engines read facts like your prices, opening hours and product details, and it makes clear which brand a page is about. That can help a platform tell you apart from another business with your name. Just don't expect it to make up for a thin page.
For example, Harrow & Pine adds Product markup to all 200 of its product pages, and the next month its citations haven't moved. The pages that do get cited are the ones with clear descriptions, real dimensions and honest care instructions in the text itself.
In AnswerOrbit
Page checks in Recommendations look at structured data as one of eight checks, next to whether the page shows without scripts, whether it has a visible date and whether its opening answers the prompt.