An analysis by AIOGeoScan of 1,233 diagnostic records from July 2026 found that structured data in JSON-LD was unimplemented on 41.9% of pages, while 120 cases of duplicate definitions using the same @type on a single page were detected.
A survey using the AIOGeoScan diagnostic platform provided by 株式会社Bennu analyzed 1,233 diagnostic logs collected in July 2026 and found that structured data in JSON-LD was unimplemented on 41.9% of pages (516 pages). In addition, 120 cases were detected in which sites with structured data implemented had duplicate definitions of the same @type, such as Organization and WebSite, in different script tags on a single page.
The survey was conducted to shed light on the state of SEO and AIO/LLMO readiness in the era of AI search. It highlighted not only implementation rates for structured data but also qualitative issues. The findings pointed to the risk of inaccurate interpretation by AI crawlers when different JSON-LD blocks are placed in shared headers and individual templates, resulting in multiple definitions being mixed together.
The three key findings were, first, interpretation loss caused by duplicate JSON-LD definitions; second, the absence of E-E-A-T reliability signals on 70.9% of pages; and third, a 78.1% deficiency rate in answer-target structures, including summaries immediately after headings. The rates at which llms.txt and robots.txt had not been installed were also identified as issues.