How AI answers pick sources · Official documentation
Google, Bing and a controlled test give three different answers on schema and AI citations
It depends on whose document you read. Google says structured data isn't required for generative AI search; Bing says structured content helps AI systems interpret and summarize pages. In an Ahrefs test comparing pages that newly added schema with matched controls, citation changes were -4.6% in Google AI Overviews, +2.4% in Google AI Mode, and +2.2% in ChatGPT.
"Add schema and AI will cite you" is one of the most repeated lines in GEO pitches. The opposite claim, that AI search ignores schema, circulates just as widely.
We traced both back to three primary documents: Google's guide, a Bing Webmaster Blog post, and an experiment that compared pages newly adding schema with matched controls.
Finding 01
Google and Bing describe schema's role in AI search differently
Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add. However, it's a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search.
Structured content, like schema-marked product pages, FAQs, and comparison tables, helps AI systems interpret and summarize your content more effectively.
Google's sentence is about generative features in Google Search, and it still recommends schema for rich result eligibility. Bing's sentence is about AI summaries in Bing and Copilot. The same markup carries a different expectation depending on the engine in question.
Finding 02
In the experiment, Google AI Overviews citations fell 4.6% relative to matched controls
Adding schema produced no major uplift in citations on any platform.
Notes are the experiment's own verdicts, verbatim. Source: Ahrefs, "We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved." (2026-05-11).
The test measured citation changes on pages that added schema. Whether schema helps AI systems find pages they cannot yet read is, by the experiment's own account, outside its scope.
So what is schema actually for?
- Add it to pages that need Google rich result eligibility, and write that purpose into the proposal.
- If Bing and Copilot are in scope, pair it with structured tables and product information.
- Put the material for citation in the visible text: definitions, figures and comparisons.
Methodology
- Google Search Central, "Optimizing your website for generative AI features on Google Search" (updated 2026-07-10) · checked 2026-09-13 · https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Bing Webmaster Blog, "How AI Search Is Changing the Way Conversions are Measured" (updated 2025-11-20) · checked 2026-09-13 · https://blogs.bing.com/webmaster/November-2025/How-AI-Search-Is-Changing%E2%80%AFthe%E2%80%AFWay%E2%80%AFConversions%E2%80%AFare-Measured
- Ahrefs, "We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved." (updated 2026-05-11) · checked 2026-09-13 · https://ahrefs.com/blog/schema-ai-citations/
- Experiment design: 1,885 treated pages and 4,000 control pages, matched difference-in-differences over 30 days before and after. The method is published.
- Limits of the experiment: the sample leaned toward pages that already had many AI citations, so results may not carry over to new sites. Only the AI Overviews decline was statistically significant; the other two are indistinguishable from zero.
- Bing's sentence is one passage in a post about measuring conversions. The post gives no figure that isolates schema.