TL;DR: Google's own documentation says no special structured data is needed to appear in AI Overviews, and the FAQ rich result has been restricted to government and health sites since 2023. Schema markup is still worth writing, but the reason changed: it is machine comprehension infrastructure now, not a button that buys search features.
The one-sentence answer: Add structured data to make your pages unambiguous to machines, keep your business entity data accurate everywhere it appears, and stop expecting any specific visual reward for it.
Somewhere right now, an agency is selling a small business "schema optimization for AI search" with the promise that the right markup gets you cited by AI.
Google has answered that pitch in writing. The honest version of this topic is more interesting than the folklore, so here is what the primary documents actually say, and what we still mark up on every page anyway.
What does Google actually say schema does for AI features?
The bluntest sentence in this whole topic sits in Google's documentation on AI features and websites: "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add."
The same document states there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary," and tells site owners to "apply the same foundational SEO best practices for AI features as you do for Google Search overall."
That is the primary source. Any pitch that promises schema markup as the entry ticket to AI Overviews is contradicted by the company running AI Overviews.
So is structured data dead?
No, and Google's other documentation explains why in one line: structured data provides "explicit clues about the meaning of a page," and "Google uses structured data that it finds on the web to understand the content of the page."
Understanding is the durable product. A page that declares in JSON-LD that it is an Article, written by this person, published on this date, about this topic, for this organization, at this address, has removed a layer of guesswork for every machine that reads it. Rich results were always just the visible reward for that clarity. The clarity itself did not stop mattering when the rewards got rarer.
What happened to FAQ and HowTo rich results?
The cautionary tale of this whole subject. For years, FAQPage markup reliably earned expandable Q&A dropdowns in Google results, so the industry stamped FAQ blocks on everything. Then Google pulled the feature back in stages, documented in its own changelog: a September 2023 restriction limiting the rich result to "well-known, authoritative government and health websites," a May 2026 announcement that the FAQ rich result "is no longer shown in Google Search results" at all, and removal of the feature's documentation in June 2026.
For everyone selling schema as a rich-result vending machine, the product was discontinued. The lesson is about what schema is for: any strategy built on a specific visual treatment is renting space in an interface Google can redecorate whenever it likes.
Why do we still emit FAQPage on every post?
Because the markup's meaning outlived its rich result, and the cost of shipping it is close to zero.
This blog emits BlogPosting JSON-LD on every post and FAQPage markup on the 99 of 101 posts that carry a question-and-answer section, all generated automatically from the content itself. We kept that pipeline after the rich result went away, and through May 2026 this site's articles accumulated 932 Bing AI citations on a single post and a sitewide peak of 171 AI citations per day, with our question-formatted, capsule-structured content being what the AI systems cited.
The honest attribution: we cannot isolate schema's contribution from the content structure it mirrors, and we do not claim the markup alone earned a single citation. What we can say from our own operation is that structuring content as explicit questions with self-contained answers, and declaring that structure in markup, is one integrated practice, and that practice measurably got our pages into AI answers while our Google clicks were still near zero.
Which structured data types are still worth writing?
The ones that describe things machines need to identify, rather than decorations machines used to draw.
| Type | Why it still earns its bytes |
|---|---|
| Article / BlogPosting | Author, date, publisher, headline: provenance signals machines check |
| Organization | Who is behind the site; ties pages to one entity |
| LocalBusiness | Address, hours, geo, departments; feeds knowledge panels and business info |
| Product | Price, availability, reviews for commerce queries |
| Breadcrumb | Site structure, cheap and unambiguous |
| FAQPage | Meaning intact for machine readers; just expect no visual feature |
LocalBusiness deserves the emphasis for service companies: Google's documentation says the knowledge panel with business details is what this markup feeds ("you can tell Google about business hours, different departments within a business, reviews"), and Google's AI-features best practices separately include keeping Business Profile information up-to-date. That pairing is as close as the primary sources get to connecting markup with AI-era visibility, and it runs through your data being correct, everywhere.
How do you implement structured data without hiring a developer?
Three decisions, and the first two are already made for you.
Format: Google supports JSON-LD, Microdata, and RDFa, and recommends JSON-LD, which is also the only one of the three that lives in a single self-contained script block instead of being threaded through your page's HTML. Use JSON-LD.
Completeness: Google's documentation is explicit that "you must include all the required properties for an object to be eligible for appearance in Google Search with enhanced display." Half-filled markup is not partial credit; a LocalBusiness block missing its address is a LocalBusiness block that does nothing. Fill every required property or pick a simpler type you can complete.
Generation: the sustainable version is markup produced automatically from your content, never maintained by hand. On this site, every post's FAQPage and BlogPosting JSON-LD is generated from the article's own headings and question sections at build time, which is why keeping it costs us nothing and why it never drifts out of sync with the page. If your platform can't do that, most mainstream CMS ecosystems have plugins that generate the core types from templates; the thing to verify is that what they emit matches what the page visibly says, because markup contradicting visible content is worse than no markup.
What should you do instead of chasing rich results?
Get the eligibility layer right, because that one is documented. Google states that content becomes eligible for AI features by being "indexed and eligible to be shown in Google Search with a snippet." Indexing, snippet eligibility, and crawler access are the actual gate. In our site audits, the most common silent AI-visibility killer is not missing schema; it is CDN and security-plugin "block AI bots" toggles quietly switched on, cutting sites out of AI answers without the owner ever choosing that. We covered that check, and eight others, in the AI SEO audit checklist.
Then spend the effort schema-sellers want for markup on the thing the machines actually quote: content structured for extraction. That is answer engine optimization, and it is where the visible results in our data came from.
Frequently Asked Questions
Does schema markup help with ChatGPT or Perplexity citations?
Neither company documents structured data as a citation factor, so any confident claim in either direction is speculation. What is documented: schema.org is the web's shared vocabulary for declaring what a page is, and machine readers of every kind benefit from unambiguous pages. Treat it as good hygiene with unknown upside there, not as a lever with a promised payoff.
Should we remove FAQ markup that no longer produces rich results?
We kept ours, and none of Google's changes required removal; unused structured data simply goes unrewarded visually while its meaning stays readable to every machine that parses the page. Removal only makes sense if the markup is maintained by hand and costs real effort. If it is generated from your content automatically, it is free clarity.
Is JSON-LD really better than the other formats?
It is the format Google recommends ("in most cases, that's JSON-LD"), and it is the easiest to generate programmatically because it lives in one script block instead of being woven through your HTML. Every consideration that matters for a small business site points the same direction.
The audit we run on client sites checks the whole eligibility chain in order: indexing, snippet eligibility, crawler access, then structure and markup, because that is the order the primary sources support. If your current provider is selling markup as an AI-visibility shortcut, the checklist above is what to hold it against.
See ClickWerxs AI SEO or get in touch.
Sources
- Google Search Central, "AI features and your website" — no special structured data or markup required for AI Overviews/AI Mode; eligibility via indexing and snippet eligibility; business information guidance. developers.google.com (retrieved 2026-08-13)
- Google Search Central, "Intro to structured data" — structured data as explicit clues for page meaning; eligibility not guarantee; JSON-LD recommendation. developers.google.com (retrieved 2026-08-13)
- Google Search Central changelog, FAQPage structured data — September 2023 restriction to "well-known, authoritative government and health websites"; May 2026 announcement that the FAQ rich result is no longer shown in Google Search results; June 2026 documentation removal. developers.google.com (retrieved 2026-08-13)
- Google Search Central, LocalBusiness structured data documentation — knowledge panel display, hours, departments, reviews. developers.google.com (retrieved 2026-08-13)
ClickWerxs sells AI SEO services, including the audit practices described in this post. First-party citation figures are from our own analytics and are a past result, not a promise of outcomes for any engagement.
Kaleb Dickhaut — Founder, ClickWerxs. Kaleb built ClickWerxs from the ground up, from payment processing ISO to the Command Center platform to the AI SEO methodology the blog runs on. He has onboarded hundreds of small businesses onto payment and CRM systems. linkedin.com/in/kaleb-dickhaut