The Schema.org vocabulary now spans more than 800 different types, and plenty of websites overdo it: they tag everything they can find, including deprecated or irrelevant types, while neglecting the handful of properties that genuinely move rankings and visibility inside AI-generated search results. In 2026, the “more markup is always better” approach no longer holds. Google has tightened its criteria, retired entire rich result features, and leaned harder on structured data as a trust signal for its generative AI systems.
What changed with structured data in 2026
Three major shifts happened early in the year, and every site owner should know about them:
- Several types deprecated in January 2026: Practice Problem, Dataset, and Sitelinks Search Box were removed from Google Search’s supported list, making that markup pointless for rich results going forward.
- Large-scale FAQPage demotion in March 2026: an update sharply reduced FAQ snippet display in search results, particularly when the markup didn’t exactly match the visible page content.
- FAQ rich result retired on May 7, 2026: Google officially removed this visual search feature. FAQPage markup is still technically valid and still helps Google understand page content, but it no longer produces the enhanced listing that used to win extra SERP real estate.
The direct consequence: a markup strategy designed back in 2023 or 2024 is likely outdated on several fronts, and some JSON-LD scripts are now running for nothing.
The 5 schema types that actually matter in 2026
1. Organization
This is the foundation of any structured data strategy. Complete Organization markup — name, logo, social profiles, contact details, sameAs identifiers — helps Google establish your site as a recognized entity in its Knowledge Graph. It’s one of the most underrated levers for getting cited by AI answer engines today.
2. Article / BlogPosting
For any editorial site or company blog, this markup remains essential: it conveys publish date, author, and lead image — signals that generative AI systems actively use to judge freshness and reliability before citing a piece of content.
3. FAQPage, with caveats
Even without a visual rich result since May 2026, FAQPage markup still has a job to do: it helps Google extract direct answers for AI Overviews. The rule is strict: every marked-up question and answer must match the visible page content word for word, or risk being ignored — or penalized.
4. Product
For e-commerce sites, this remains one of the most profitable types: price, availability, reviews, and ratings still generate high-impact rich results for click-through rate, and feed directly into comparison engines and shopping results.
5. LocalBusiness
For local businesses and service providers, this markup is still the technical bridge between your website and Google’s local ecosystem — Google Business Profile, Maps, “near me” searches. It must stay perfectly consistent with the NAP details published elsewhere.
Why markup is becoming critical in the generative search era
AI Overviews now appear on 50 to 60% of US searches, powered since late January 2026 by more capable language models. These systems don’t “read” a page the way a human does — they rely heavily on structured data to decide which sources to cite and how to surface information reliably. A site with consistent, complete Organization, Article, and Person markup has a statistically much higher chance of being pulled into an AI-generated answer than an identical site with no structured data at all.
The single most powerful lever, often overlooked, isn’t even visible on the results page: entity markup that precisely identifies your organization in the Knowledge Graph. It produces no visual rich result, but it directly influences your odds of being cited in AI mode and the accuracy of your knowledge panel.
Implementation best practices to follow
- Stick exclusively to JSON-LD, placed in the page head — still Google’s recommended format and the easiest to maintain long-term.
- Fill in relevant optional properties: incomplete markup can outright suppress a rich result, even when every required property is present.
- Never mark up information that isn’t visible on the page. Schema’s job is to clarify existing content, not to add claims a visitor can’t verify on screen.
- Check consistency across every page: an Organization entry that’s accurate on one page and wrong on another sends a degraded trust signal.
Habits worth dropping in 2026
Three carried-over habits are now actively hurting the sites that keep them: tagging FAQPage hoping to recover a rich result that no longer exists, maintaining deprecated types like Dataset or Practice Problem with zero real benefit, and stacking generic schema on every page without checking relevance. Even a quick markup audit usually uncovers these dead JSON-LD scripts and clears the way to implement the five types that genuinely matter.
Want to know if your site is still running outdated schema types, or missing the markup that matters for generative AI search? Get a full SEO audit from seoforge.io and walk away with a concrete action plan for your markup, indexation, and visibility in AI-driven results.
