Schema markup for businesses: what machines need to read.

A language model reading your homepage has to infer what kind of business you are from marketing copy — copy written to evoke, not to specify. Schema removes the inference. When an engine can read your entity data directly, it doesn't have to guess — and engines that don't guess don't hallucinate your specialties, merge you with the business two blocks over, or hedge instead of naming you.

This is live-parse-layer work: Google AI Overviews and Copilot read it on every crawl, which is why schema fixes are usually the first thing to show results in an engagement — weeks, not quarters.

Organization — or LocalBusiness, if you serve a physical market — goes on the homepage and establishes the entity: name, address, phone, hours, the works. Person goes on every team bio, because buyers ask engines about people by name. Service goes on each service page, so engines can match you to service-level queries instead of just category ones. AggregateRating and Review markup belongs wherever real reviews live — honestly marked up, never fabricated. And FAQPage belongs wherever you answer real buyer questions, because question-formatted content maps directly onto how buyers prompt engines.

Every block must agree exactly with your Google Business Profile and with every directory listing — same name, same address, same categories. Schema that contradicts the rest of your record doesn't strengthen the entity; it splits it.

Most businesses that have schema got it from a plugin or an old agency, and it's working against them. The recurring offenders: a homepage typed as a generic Organization when LocalBusiness would win the local queries; schema that disagrees with the Google Business Profile on hours or address; star ratings marked up with no visible reviews behind them; and duplicate blocks from two plugins describing what reads, to a machine, like two different businesses.

Run your homepage and one service page through Google's Rich Results Test. No detectable structured data, or errors on the basics? That's week-one work in any engagement — yours or ours.

Schema lives as a block of JSON-LD in the head of a page — a format you can read with no programming background, which matters because you should read yours. Whoever maintains your site can paste a well-formed block in minutes; the hard part was never the code, it's knowing what the block should say. Practical guidance: write the block from your real entity data, validate it in Google's Rich Results Test before and after publishing, and re-check after every site update — plugins overwrite things silently.

Budget reality: this is hours of work, not weeks. In our rebuild phase the full schema pass fits inside the 14-day infrastructure window alongside the knowledge panel and directory work — it's the cheapest leverage in the entire engagement.

Schema alone won't get you named — it's one layer of three, and the citation work in trusted sources is what makes engines want to mention you at all. But schema is the layer that makes every other signal attach to the right entity. Citations pointing at a business whose own data is contradictory are citations wasted. Foundation first; it's cheap, it's fast, and it's entirely under your control.