Schema Markup Implementation Guide
Schema markup stopped being just a rich-snippet trick. In 2026 it is how Google, ChatGPT, and Perplexity verify what your business is and what it sells before they decide whether to cite you. This guide walks through the four schema types most service and software businesses need, with copy-ready JSON-LD and a validation workflow that catches the mistakes that silently kill rich results.
In this article7
- What schema markup actually is (and the only format worth using)
- Why schema matters more in 2026 than it did for rich snippets
- LocalBusiness schema for a physical or service-area business
- Product schema for pages where people can buy
- FAQPage and Article schema, post-2026 reality
- The validation workflow that catches silent failures
- How this fits a real technical SEO workflow
What schema markup actually is (and the only format worth using)
Schema markup is a vocabulary from schema.org that describes the things on a page in a way machines can parse without guessing. A page about a dentist becomes a Dentist entity with an address, hours, and phone. A blog post becomes an Article with an author and a publish date. You express that vocabulary in one of three syntaxes, but only one matters in practice: JSON-LD. Google recommends JSON-LD because it sits in a single script block, separate from your HTML, so it is easier to generate, easier to maintain, and far less error-prone than Microdata or RDFa woven through your markup. You place a script tag of type application/ld+json in the head or body of the page, and that is the whole mechanism. Multiple JSON-LD blocks on one page are allowed and supported, and when you have several related entities you can nest them in a single block using an @graph array rather than scattering script tags. Do not mix Microdata and JSON-LD for the same fields. Competing declarations reduce machine trust and make debugging miserable.
Why schema matters more in 2026 than it did for rich snippets
For most of the last decade schema earned its keep by triggering visual SERP features: star ratings, FAQ accordions, recipe cards. That role narrowed sharply. Google retired FAQ rich results on May 7, 2026, and the March 2026 core update pulled back rich-result display for FAQ, Review, and How-To markup on pages where it was bolted on as a SERP-manipulation tactic. If that were the whole story, schema would be losing value. It is not. The value moved. Both Google and Microsoft confirmed in 2026 that they use schema markup to power generative AI features, and ChatGPT confirmed it uses structured data to decide which products surface in its results. When an AI system synthesizes an answer, it prefers facts it can extract with certainty. Clean JSON-LD gives it that certainty, while unstructured prose forces inference, which introduces error. Third-party studies put pages with thorough schema somewhere in the range of 35-50% more likely to be cited in AI-generated summaries. Treat those numbers as directional, since most come from vendors selling schema services and Google has not confirmed any specific lift, but the direction is consistent across sources: structured pages get cited, vague pages get skipped.
LocalBusiness schema for a physical or service-area business
If you have a real location or a defined service area, LocalBusiness is your highest-value local signal. Google's hard requirement is small: name and a full address. Everything else is recommended but is what actually makes you competitive in the Map Pack. The address must be a nested PostalAddress object with streetAddress, addressLocality, addressRegion, postalCode, and addressCountry, and that country has to be the two-letter ISO code (US, not United States). Add telephone to complete your NAP (name, address, phone), and match it character for character against your Google Business Profile. '123 Main St' in schema against '123 Main Street' on your profile is treated as an inconsistency that erodes trust. Add geo coordinates as a GeoCoordinates object with latitude and longitude at five or more decimal places (right-click your spot in Google Maps to copy them). Use the most specific subtype you can: Plumber, Dentist, or DaySpa rather than the generic LocalBusiness, because the specific type unlocks industry features. For opening hours use 24-hour time and full ISO day names: openingHours 'Mo-Fr 09:00-17:00', never '9am-5pm'. One caveat that trips people up: an online-only business should not use LocalBusiness at all. Use Organization or OnlineStore instead.
Example skeleton: {"@context":"https://schema.org","@type":"Plumber","name":"Acme Plumbing","telephone":"+1-555-123-4567","address":{"@type":"PostalAddress","streetAddress":"123 Main St","addressLocality":"Austin","addressRegion":"TX","postalCode":"78701","addressCountry":"US"},"geo":{"@type":"GeoCoordinates","latitude":30.26715,"longitude":-97.74306},"openingHoursSpecification":[{"@type":"OpeningHoursSpecification","dayOfWeek":["Monday","Tuesday","Wednesday","Thursday","Friday"],"opens":"09:00","closes":"17:00"}]}
Product schema for pages where people can buy
Product schema turns a page into something AI shopping surfaces and Google can act on. At minimum a merchant listing needs name, image, and offers, where the Offer object carries price, priceCurrency, and availability. The single most common mistake is formatting the price: it must be a plain number string, '19.99', never '$19.99' or '1,350'. Use schema.org normalized URL values for availability (https://schema.org/InStock) rather than free text. For full merchant-listing eligibility, add at least one identifier (GTIN, MPN, or SKU), brand, and aggregateRating, and add shippingDetails and hasMerchantReturnPolicy so shoppers see costs and return terms. Pages with complete shipping and return data tend to get preferential treatment, because Google rewards transparency. Two non-obvious rules: there must be a single offer on the landing page for Google to match it cleanly to your product data, and you should never add an aggregateRating unless that rating is genuinely shown on the page and substantiated. When the visible price changes, the JSON-LD has to change at the same moment. A stale price in schema is the kind of mismatch that gets markup ignored or, increasingly, flagged.
FAQPage and Article schema, post-2026 reality
FAQPage is the schema type whose role changed the most. Since Google retired FAQ rich results in May 2026, FAQPage markup no longer produces the accordion chips in search. It is not useless, though: it still cleanly structures question-and-answer pairs that AI systems extract as direct answers, so keep it on pages where the Q and A genuinely matches visible content, and drop it from pages where it was only there to grab SERP real estate. Article schema (use the specific subtype: BlogPosting for blogs, NewsArticle for time-sensitive news) has no strictly required fields, but four carry almost all the weight: headline, image, datePublished, and author. Keep headline under 110 characters and make it match your H1, since a mismatch gets flagged. Provide images at 1200px or wider, ideally in 1:1, 4:3, and 16:9 ratios so you qualify across Discover and Top Stories surfaces. Critically, author must be a Person or Organization object with a name and a url to an author page, not a bare string. 'author':'Jane Smith' is the version AI extraction systems ignore. Use ISO 8601 dates ('2026-06-22' or with a time), and keep dateModified honest, because Google reads it as a freshness signal. If you update an article and leave dateModified stale, Google may treat it as never updated.
The validation workflow that catches silent failures
Schema that validates in a generic checker can still fail Google's stricter eligibility rules, so use the right tool for each stage. Before you publish, run the markup through Google's Rich Results Test, which is the only validator whose 'eligible for X' verdict matches what Google's index will accept and which previews how the result may appear. Cross-check the same markup against the schema.org Schema Markup Validator (the successor to the old Structured Data Testing Tool) for full spec compliance, which matters for schema types Google does not support as rich results. After deploy, retest the live URL and then watch the Enhancements reports in Google Search Console over the following two weeks, since that is where real errors, impacted URLs, and actual SERP-feature appearances show up across the whole site. The non-negotiable rule underneath all of it: the schema must describe content that is actually visible to a human on the page. Google penalizes invisible or mismatched markup rather than just ignoring it, and AI systems are moving toward cross-referencing schema claims against the live page. Watch your JSON syntax too, because one stray comma or unbalanced brace means the entire block is silently discarded.
How this fits a real technical SEO workflow
Schema is not a one-time copy-paste. Prices change, hours change, articles get updated, and every one of those changes has to flow into the JSON-LD at the same time the visible page changes, or the markup turns into a liability. That is why schema works best generated from your data layer rather than hand-edited per page: a single template that pulls name, address, price, author, and dates from the same source that renders the page, so they can never drift apart. This is exactly the kind of work we handle in a Pineido technical SEO sprint. We audit which schema types you are genuinely eligible for, implement complete and consistent JSON-LD wired to your content source, add the entity markup (Organization, sameAs to authoritative profiles, knowsAbout) that helps AI systems resolve who you are, and hand back a validation and monitoring setup so it stays correct. It is fixed scope, fixed price, and 30 days, which suits structured data well because the work is concrete and measurable rather than open-ended.
Frequently asked questions
Which schema format should I use: JSON-LD, Microdata, or RDFa?
Use JSON-LD. Google recommends it because it lives in a single script block of type application/ld+json, separate from your HTML, which makes it easier to generate and maintain and far less error-prone. Microdata and RDFa are woven through your markup and are harder to keep consistent. Avoid mixing formats for the same fields, since competing declarations reduce machine trust.
Does schema markup directly improve my rankings?
No, schema is not a direct ranking factor and Google has stated as much. Its value is indirect: it makes pages eligible for supported rich results, improves how search engines understand your content, and increasingly helps AI systems extract and cite your facts with confidence. Those effects compound over time, but correct markup alone never guarantees a higher rank or a rich result.
Is FAQPage schema still worth adding after the 2026 changes?
It depends on the goal. Google retired FAQ rich results on May 7, 2026, so FAQPage no longer produces accordion chips in search. It is still useful for structuring genuine question-and-answer content that AI systems extract as direct answers. Keep it where the Q and A matches visible content, and remove it from pages where it was only there to grab SERP space, since that pattern was a target of the March 2026 update.
What happens if my schema does not match what is on the page?
Mismatched or invisible schema is now actively risky rather than just ignored. Google penalizes markup that describes content not visible to human readers, and AI systems are moving toward cross-referencing schema claims against the live page. Always make sure your name, address, prices, headlines, and dates in JSON-LD match exactly what renders on the page, and update the markup the moment the visible content changes.
How do I validate schema markup before publishing?
Run the markup through Google's Rich Results Test first, since it is the only validator whose eligibility verdict matches what Google's index accepts, and it previews the result. Cross-check it against the schema.org Schema Markup Validator for full spec compliance. After deploy, retest the live URL and monitor the Enhancements reports in Google Search Console over the next two weeks for real errors and feature appearances.
What are the minimum required properties for LocalBusiness and Product schema?
For LocalBusiness, Google strictly requires only name and a full nested PostalAddress (with the country as a two-letter ISO code like US). For a Product merchant listing, you need name, image, and an offers object containing price (a plain number like 19.99), priceCurrency, and availability as a schema.org URL value. Everything beyond those, like geo coordinates, phone, identifiers, and ratings, is recommended and is what makes you competitive.
Get operations insights weekly
Join service-business operators who get one actionable insight on ops, automation, and growth every week.
No spam. Unsubscribe anytime.
You’re subscribed! Look for our next issue in your inbox.
That did not go through. Please try again, or email [email protected].
Ready to ship?
Tell us where you are stuck. We will review it and send a focused recommendation within one business day.