Emerging Digital PartnersGet in touch

Article · structured-data

The schema problem: most sites have none, and the rest have it wrong

Structured data is sold as a cheat code for AI search. It is not, and the evidence says so. What it actually does is stop engines misidentifying your business and misreading your pages, and on most sites we audit it is either absent or actively wrong. Here is what we find and what to do about it.

By Richard Daniel3 min read

Two claims about structured data circulate at the same time. The first is that JSON-LD is the secret to AI citations: add FAQ schema and HowTo schema and the engines will queue up. The second is that Google has said no special markup is needed for AI Overviews, so schema is a waste of time. Both are wrong, and both lead to the same outcome, which is sites that either have no schema or have schema nobody has looked at.

What the evidence actually says

SE Ranking analysed pages cited by ChatGPT and found that around 71 percent carried structured data, with a similar figure for Google AI Mode. That is a correlation: well-run sites tend to have schema and tend to get cited. Ahrefs then ran the experiment that matters. In May 2026 they added JSON-LD to 1,885 pages that were already being cited and measured no change in citation rates. Adding markup to a page that already works does not make it work more.

Google's May 2026 guide is consistent with that: no special markup, markdown or files are required to appear in its AI features. The same guide also says the SEO fundamentals still apply, and structured data for Organization, Article, Product and LocalBusiness has been part of those fundamentals for a decade.

So the honest position is this. Schema is not a lever that raises citations. It is a way of stating facts in a form machines cannot misread: this page is an article by this person, published by this organisation, at this address, offering this service at this price. Its job is to prevent misidentification. On the sites we audit, misidentification is a real and common failure, so the job is worth doing.

The five failures we see most

1. Nothing. Roughly half the small business sites we look at have no structured data at all beyond whatever the theme emits for the page title. The engine has to infer who you are from prose, and prose is where the vague-copy problem lives.

2. Two of everything. An SEO plugin emits an Organization block. A theme emits another. A reviews app emits a LocalBusiness. Three names, two addresses, one phone number that changed in 2023. The engine now has a conflict and no way to resolve it. This is worse than nothing.

3. Placeholders. "Company Name", "Your Address Here", a logo URL that returns 404, "price": "0.00". Generated on install, never edited. Validators pass it because the syntax is fine; a human reading it would wince.

4. FAQ schema for invisible content. Questions marked up that do not appear on the page, or that appear only inside a JavaScript accordion that renders empty in the HTML. Google's guidelines require markup to describe visible content; engines treat the mismatch as a trust signal in the wrong direction.

5. Dates that lie. dateModified rewritten on every page load by a plugin, or a datePublished of 1970. Engines use dates to judge freshness; a dishonest date does more harm than a missing one.

What to do, in order

Start with one Organization or LocalBusiness block on the homepage with a stable @id, the same 25-word description you use everywhere, real address and phone, founding date and sameAs links to LinkedIn, Google Business Profile and Companies House. Remove every duplicate block; if two plugins fight, disable schema in one.

Then Article on every piece of content with a real author, real dates and publisher pointing at the Organization @id. Then Service or Product on the pages that sell, with prices that match the visible page and the Merchant Center feed. Then BreadcrumbList. Then FAQPage only where the questions are visible.

Validate every template in the Schema.org validator and Google's Rich Results Test, and then read the output as a person. The validator does not know your company name is wrong.

The point

The businesses that appear consistently in AI answers are not there because of schema. They are there because the engines are confident about who they are and what they offer, and schema is one of the few places you can state that in a form that cannot be misread. Treat it as a fact sheet, keep it consistent with the page, and stop expecting it to do the content's job.

Frequently asked questions

Does adding schema get you cited by AI?

+

Not on its own. SE Ranking found about 71 percent of pages cited by ChatGPT carried structured data, which is correlation. Ahrefs' May 2026 experiment added JSON-LD to pages that were already cited and found no measurable lift. Schema's value is disambiguation: making sure engines identify you correctly and understand what each page is.

What is the most damaging schema mistake?

+

Conflicting Organization blocks, usually from two plugins, that give the engine two different company names or addresses for the same site. Second is FAQPage markup for questions that are not on the page, which engines and Google both treat as a trust problem.

Sources

  1. ReachRoller: What ChatGPT actually cites (SE Ranking and Ahrefs findings) · reachroller.com
  2. Google Search Central: Structured data general guidelines · developers.google.com
  3. Schema.org validator · validator.schema.org

Richard Daniel

Automation and Delivery Lead, Emerging Group

Richard leads automation and delivery across the Emerging group, working with EDP on client websites and with ETT on enterprise AI and process automation. He is the person who turns an audit finding into a working fix: crawler access, rendering, tracking, structured data and the plumbing that most marketing teams never see. He writes the technical guides on this site.

Article15 Sept 2026 · 3 min read

Nobody knows who you are: the entity problem behind most AI invisibility

AI engines do not rank your pages. They build a model of your business from everything they can find and decide how confident they are in it. When your name, description and facts differ from platform to platform, that confidence collapses, and an engine only recommends what it is confident about.

George McKenna
ai-searchstrategystructured-data
Article15 Sept 2026 · 3 min read

You probably blocked the AI crawlers. Most sites did it by accident.

Between Cloudflare's default settings, security plugins with a 'block AI bots' toggle, and robots.txt block lists copied from forums in 2024, a large share of business websites have opted out of AI search without anyone deciding to. Here is how it happens and how to tell if it happened to you.

Richard Daniel
technicalai-search
Article15 Sept 2026 · 4 min read

Your site ranks on Google and is invisible to AI. Here is why.

The most common AEO failure we find has nothing to do with content. Google renders JavaScript; the AI crawlers do not. A page can sit at position one in Search Console and be a blank shell to ChatGPT, Perplexity and Claude, and nobody notices because every dashboard is green.

Richard Daniel
technicalai-search

← All articles