Schema markup for small businesses — the practical 2026 guide
Schema markup is structured data — written in JSON-LD — that tells Google and AI assistants exactly what a page is about. For a UK small business in 2026, the minimum viable schema stack is five types: Organization (or LocalBusiness for premises-based businesses), Service for each commercial offering, FAQPage on any page with Q&A, BreadcrumbList for navigation, and Article on blog posts. Implementing these correctly typically lifts both Google rich-result eligibility and AI citation likelihood.
Why schema matters more in the AI era
In the 10-blue-links era, schema mostly bought you rich snippets — review stars, FAQ accordions, breadcrumb trails. Useful, but not transformational.
In the AI era it does something more important: it gives ChatGPT, Perplexity, Claude and Google AI Overviews unambiguous facts they can quote with confidence. An AI engine that knows your priceRange is '£300+', your areaServed is 'United Kingdom' and your telephone is '+44…' will recommend you far more readily than one that has to infer those facts from prose.
The five JSON-LD types every SMB needs
1. Organization (or LocalBusiness) — name, url, logo, sameAs links to social profiles, contactPoint with telephone and email. Use LocalBusiness if you serve customers from a physical address.
2. Service — one entry per commercial offering, with name, description, provider, areaServed and priceRange. This is the schema AI assistants use most heavily when matching businesses to user needs.
3. FAQPage — wraps any FAQ block. Each Question has an acceptedAnswer with the response in plain text. This is the single highest-leverage schema for AI citation.
4. BreadcrumbList — improves Google rich results and helps AI understand site hierarchy.
5. Article — on blog posts. Include headline, datePublished, dateModified and author. AI engines prefer recently-modified content.
Implementation tips that prevent common errors
Use a JSON-LD @graph wrapper to combine multiple schema types on one page without duplicating @context. Always validate output through the Schema.org validator and Google Rich Results Test before publishing. Keep your Organization schema in a single canonical location (typically the homepage) and reference it from other pages by @id rather than duplicating.
Avoid the temptation to mark up content that doesn't exist on the visible page — Google penalises this and AI assistants ignore it. Schema must reflect what the user actually sees.