#What structured data actually does
Structured data is a machine-readable description of what a page is about. It does not change your content, and it is not a direct ranking factor. What it does is make your page eligible for enhanced search features that plain HTML cannot trigger — star ratings, prices, breadcrumb trails, event dates, recipe cards, video thumbnails and site-name display.
The measurable benefit is click-through rate. Pages that win a rich result commonly see CTR improvements of 15% to 30% at the same average position. More clicks at the same rank means more traffic, and the engagement signal does you no harm either.
There is a second, growing reason: AI search systems and assistants parse structured data to understand entities and relationships. Clean schema is increasingly how you are correctly represented in generated answers.
#Use JSON-LD, not microdata
Google explicitly recommends JSON-LD, and for good practical reasons:
- It sits in a single
<script>block, separate from your visible markup - It survives template refactors that would break attribute-based microdata
- It can be generated server-side from the same data that renders the page
- It is trivially auditable — you can read it in one place
Microdata and RDFa require weaving attributes through your HTML, where they get lost the first time a designer changes a component. Unless you are maintaining a legacy implementation, there is no argument for them.
#Where to put it
Inside <head> is conventional, safest and easiest to audit. Google can also read JSON-LD from <body> and from script injected after page load, but head placement removes any dependence on rendering and makes the markup easy to find during a technical review.
One schema block per page describing the primary entity is the norm. If a page genuinely has several — an article that also contains an FAQ section and sits in a breadcrumb trail — use a @graph array or multiple script blocks. Do not describe an entity that is not on the page.
#Required versus recommended properties
Google distinguishes between required and recommended properties. Missing a required property makes the page ineligible for the rich result entirely — there is no partial credit. Missing a recommended one usually just means a less complete display.
This generator marks required fields and refuses to report the output as valid until they are present. That single check catches the most common structured data failure, which is an Article missing datePublished or a Product missing offers.
#Which type to use
- BlogPosting / Article — blog posts, news, guides. The default for editorial content.
- Product — anything with a price and availability. Include
offersand, where genuine,aggregateRating. - FAQPage — a page with visible question-and-answer content.
- LocalBusiness — a physical location with an address and opening hours. Use the most specific subtype available for your sector.
- Organization — your homepage or about page, establishing the brand entity.
- Person — author pages, which help associate content with a named expert.
- Event — anything with a date, time and location.
- Recipe, VideoObject, HowTo, SoftwareApplication — self-explanatory, each with its own required set.
- BreadcrumbList — on every page in a hierarchy. Cheap to add and reliably displayed.
#The rules that get sites penalised
Structured data is subject to spam policies, and violations can trigger manual actions that remove all rich results from a site.
Mark up only what is visible on the page. Rating markup for reviews that do not appear, prices that differ from the page, or FAQ content hidden from users are all violations.
Do not invent ratings. Self-serving aggregateRating on your own product or service page, with no visible review content, is explicitly against policy and is actively enforced.
Be specific and accurate. An Organization claiming a foundingDate you made up, or a Product with an availability that never changes, degrades trust in your markup generally.
FAQ rich results are largely gone. Google restricted them to authoritative government and health sites in 2023. Valid FAQPage markup still helps machines parse the page and costs nothing, but do not build a content strategy expecting the visual snippet.
#Validate, deploy, then monitor
- Paste the generated block into your page head
- Run the URL through the Google Rich Results Test to confirm eligibility
- Use the Schema Markup Validator at validator.schema.org for full vocabulary checking, which is stricter than Google's tool
- Watch the Enhancements reports in Search Console — errors appear there days after deployment, and a template bug will hit every page at once
That last step is the one people skip. A schema error introduced by a template change propagates silently across thousands of URLs, and Search Console is the only place it surfaces.
#Automate it
Hand-writing JSON-LD per page does not scale. Use this generator to design and validate the shape you want, then implement that shape as a template in your CMS or framework so every page of that type produces it automatically from real data. Manual markup drifts out of sync with the page within weeks.