Does adding JSON-LD schema raise AI citations? Ahrefs test

Schema is often sold as a way to get cited by AI. Ahrefs tested that claim with a before and after design instead of a raw correlation. This write-up reports that study, by Louise Linehan with Xibeijia Guan, published on 11 May 2026. SEO Madman did not run it.

What did Ahrefs test?#

The starting observation was correlational. Across 6 million URLs, AI-cited pages were almost three times more likely to have JSON-LD than non-cited pages. To test cause, Ahrefs tracked 1,885 pages that added JSON-LD between August 2025 and March 2026, each matched to 3 control URLs from other domains with similar prior citation levels. Citations were counted in the 30 days before and after the first crawl that detected schema. See the structured-data hub for related material.

Did adding schema increase AI citations?#

No, on the platforms measured. Google AI Mode (+2.4 percent) and ChatGPT (+2.2 percent) moved by amounts Ahrefs calls statistically indistinguishable from zero. Google AI Overviews fell 4.6 percent, a statistically significant drop of about 12 daily citations per page that Ahrefs says it cannot explain.

Four analyses told the same story. One was an event study that tracked citations week by week to check whether treated and control pages were drifting apart before schema was added. They tracked closely before week zero and rose together after it, which Ahrefs reads as a platform-wide AI Mode boom rather than a schema effect. Another re-ran the difference-in-differences test with a symmetrical window that excluded the recrawling period.

Why did the raw correlation mislead?#

Schema is common on pages that AI cites, but common is not causal: 53 percent of AI-cited pages have it, and those pages likely differ in authority, content and links. Ahrefs’ own reading is that if the rest of the SEO work is done well, the other signals probably carry a page to citation with or without schema. The matched-control design removes platform-wide trends but not everything else a site changes when it adds schema.

What does this not show?#

The pages already earned 100 or more AI Overview citations, so the result says little about pages AI never surfaces. Ahrefs also pooled all schema types, naming Article, FAQ, Product, HowTo and Organization, so one type could help while another does not. It measured 30 days only, and a slower effect might show in a 60 or 90 day window. It studied JSON-LD in the page HTML only, not Microdata, RDFa or schema injected with JavaScript, which Ahrefs says AI crawlers appear to treat differently.

What does Google say about schema and AI features?#

Google’s documentation is consistent with the result. Its AI features page says there are no additional requirements or special optimizations for AI Overviews or AI Mode, and that you “don’t need to create new machine readable files, AI text files, or markup to appear in these features.” It asks that structured data match the visible text on the page. This site collects the other official statements in what Google and Bing say about schema and AI.

Is schema still worth adding?#

Yes, for its documented purposes. Ahrefs lists rich results, voice assistants, knowledge graphs and downstream entity recognition as good reasons to keep JSON-LD. Google describes structured data as explicit clues about a page’s meaning and recommends JSON-LD as the easiest format to maintain, while its guidelines state it “does not guarantee that your structured data will show up in search results,” even when the markup is valid. Check each block against the required properties with the JSON-LD validator. Treat schema as a machine-readable clarity aid, not a citation lever.