Why schema matters more than ever

Generative engines do not have time to infer what your page is about. They parse structured data first, then read the prose to confirm. Pages with valid, matching schema are dramatically more likely to be selected as a source — in client tests, 3–5× more likely for FAQ-style queries.

The bar is also higher than it used to be. Schema must match the visible content. Mismatched FAQ schema (questions in markup that are not on the page) gets demoted, sometimes manually penalised. Ship schema honestly or do not ship it at all.

Article — the workhorse of every blog post

Every long-form post needs Article (or BlogPosting) schema with: headline, datePublished, dateModified, author (Person), publisher (Organization with logo), image (multiple sizes), mainEntityOfPage, articleSection, wordCount, inLanguage, and keywords.

The two fields most teams skip — datePublished and dateModified — are exactly the two AI engines weight most heavily for freshness. Ship them, and update dateModified honestly each time you revise the post.

"You cannot be cited as a source if you have not declared what kind of source you are. Schema is the declaration."

FAQPage — the citation magnet

FAQPage schema is the single highest-leverage block in your stack. AI engines lift FAQ entries directly into answer summaries because they are pre-formatted as Q&A.

Build them on every long-form post. Phrase the questions exactly as a real buyer would ask them ("How do I…", "What is the difference between…", "Is X better than Y for…"). Keep answers tight (40–80 words) and lift-able as a single sentence.

Validation note: every Question in your schema must be visible on the page in the same wording. Otherwise Google flags it.

HowTo — for any process content

If your page describes a process — "how to migrate from Shopify to BigCommerce", "how to set up GA4 for a SaaS", "how to onboard a B2B fintech customer" — it needs HowTo schema with named steps.

Each step gets a name, text, and (ideally) an image. AI Overviews surfaces HowTo content as numbered cards. ChatGPT lifts the step list into structured answers. Pages without HowTo schema on procedural queries get skipped in favour of pages that have it.

Product, Organization, Person — the entity layer

Product schema (with offers, aggregateRating, brand) is required for any commerce or SaaS product page if you want to be surfaced in product comparison answers.

Organization schema with sameAs links to your LinkedIn, Crunchbase, Wikidata, GitHub and X profiles is how you confirm to generative engines that you are a real, traceable entity. Without it, you are a string of characters; with it, you are a known node in the knowledge graph — see client implementations where this single fix moved citation share inside 60 days.

Person schema with sameAs is the same idea applied to authors. In 2026, an article authored by an unidentified person is meaningfully less likely to be cited than the same article authored by a verified Person entity.

How do you validate schema before shipping?

Three tools, in order: Schema.org validator (catches structural errors), Google Rich Results Test (catches enhancement-eligibility issues), and a manual spot check in Search Console > Enhancements after deploy.

Re-validate quarterly. Schema specs evolve, Google's eligibility rules tighten, and a markup that earned rich results last year may quietly stop earning them this year. Quarterly cadence catches the drift before it costs you traffic — or book a 14-day audit and I will deploy the full stack for you.