Schema markup is one of those topics that’s either treated as a magic ranking trick (it isn’t) or dismissed as plumbing nobody needs to think about (also wrong). The honest position: most schema does very little visibly, a handful of schema types do a great deal, and the AI search era has shifted which ones matter most.
What still doesn’t move much
Generic WebPage and WebSite schema. Article schema on basic blog posts. BreadcrumbList. All present-and-correct table stakes, useful for Google to parse, but don’t change rankings or AI citations on their own.
The four schema types that genuinely matter in 2026
Person schema with sameAs. Named authors with verifiable identity signals — LinkedIn, professional registers, Wikipedia where applicable. This is the single highest-leverage schema for E-E-A-T-sensitive sectors. AI systems weight named, verifiable expert authorship significantly more than “The Firm” byline content.
FAQPage with genuinely substantive Q&As. Not stuffed FAQs gaming rich snippets — real questions buyers actually ask, with substantive answers. This is the schema AI search systems most actively pull from when constructing answers. Get this right on your top commercial pages.
Service schema with sector-specific audience. Service schema that names exactly what you do, for whom (audienceType), where (areaServed), with a provider chain back to your Organisation. Helps Google and AI systems classify what you actually offer rather than guessing from page copy.
LocalBusiness with geo coordinates. For any firm with a physical location, full LocalBusiness schema with precise coordinates, opening hours, AggregateRating from real reviews. This is the schema Google Maps and AI assistants use to surface local options.
What’s coming
The interesting development: AI search systems are starting to favour structured data they can verify against external sources (the same way Google has done for years). Schema that claims things you can’t back up is worse than no schema. Schema that aligns with verifiable external signals — your real LinkedIn, your real Google Business Profile, your real reviews — compounds.
If your schema is patchy or hasn’t been reviewed in two years, that’s a one-month project with disproportionate return. Intro call here.
