Structured data (Schema) worth doing
TL;DR
- Structured data describes content for machines, using the Schema.org vocabulary.
- It enables rich results and helps engines and LLMs understand your site.
- Prioritize the types that apply to you; don't mark up everything for its own sake.
- Always validate in the Rich Results Test.
Structured data is markup (Schema.org) that translates your content for machines. Done well, it unlocks rich results on Google and makes your information easier for LLMs to interpret.
Which types are worth it
| Type | For what |
|---|---|
| Article | Articles and editorial content |
| FAQPage | Question-and-answer sections |
| Product / Service | Products and services |
| Person / Organization | Entity authority |
See how entity ties into GEO. I implement and validate this in technical SEO.
How to implement it: JSON-LD in the head
Google's preferred format is JSON-LD - a script block in the page head, separate from the visible HTML, so it's easy to maintain and never touches your layout. Microdata and RDFa still work, but JSON-LD is cleaner and what I use by default. One rule outranks the rest: the markup must describe what's actually on the page. Schema that contradicts the visible content is worse than none - it invites warnings and, in the worst case, a manual action.
What each type actually wins you
Article feeds Top Stories and article rich results and clarifies authorship for E-E-A-T. FAQPage can surface expandable Q&A in the SERP and hands AI clean question-answer pairs to lift. Product unlocks price, availability and review stars - among the highest-CTR rich results. Breadcrumb replaces the raw URL with a readable path. And Person/Organization with a reused @id and sameAs is the backbone of entity authority - the part that most helps you get recognized and cited.
Why it matters more in the AI era
Structured data was always about helping Google parse you. Now it also helps LLMs: explicit, machine-readable facts about who you are and what a page says are exactly what generative engines prefer to ground on. It won't fabricate authority you don't have, but where you do have it, schema makes it legible. That's why I treat Person and Organization markup as GEO infrastructure, not a nice-to-have.
A sensible implementation order
- Organization and Person site-wide, with a stable @id and sameAs.
- Breadcrumb on every non-home page.
- Article on posts; Product or Service on those pages.
- FAQPage only where a real, visible FAQ exists.
- Validate every template in the Rich Results Test before shipping.
Structured data is the cheapest way to make an honest site legible to both Google and AI - as long as it always tells the truth about the page.
Key data
| JSON-LD | 41% |
|---|---|
| WebSite | 12,7% |
| Organization | 7,2% |
| LocalBusiness | 4,0% |
The CTR uplift comes from agency studies, not Google.
Sources: HTTP Archive Web Almanac 2024
Common mistakes and how to do it right
Marking up for the sake of it causes headaches (and GSC warnings). Do it with intent:
| Avoid | Do |
|---|---|
| ✗ Marking up an FAQ that isn't visible on the page | ✓ Only marking content that is actually present and visible |
| ✗ Inventing reviews and ratings to get stars | ✓ Never fabricating - it's a manual penalty waiting to happen |
| ✗ Letting the JSON-LD drift from the content | ✓ Keeping schema and content consistent |
| ✗ Publishing without validating | ✓ Always validating in the Rich Results Test |
Tools I use and recommend
- Rich Results TestValidate structured data (free, Google).
- Screaming Frog SEO SpiderDesktop crawler for deep technical and on-page audits.
- SEO Pro ExtensionOn-page and technical checks at a glance (extension).
Sources
Frequently asked questions
Does marking up everything always help?
No. Mark what applies and is true. Misleading markup can trigger a manual penalty. See service →
How do I validate schema?
With Google's Rich Results Test and the Schema.org validator. Technical checklist →