Invisible machine-readable schema tags wrapping a small-business website

// tools in this post Google Google Gemini Gemini Perplexity Perplexity Claude Claude

We’ve said it before: most of the traffic on the internet isn’t human. It’s crawlers, indexers, and AI assistants scanning the web to decide who exists and who doesn’t.

So here’s the obvious follow-up question nobody asks: if machines are your real audience, are you writing anything for them?

Most small-business sites aren’t. They’re built entirely for human eyes — nice photos, a headline, a contact form. A person can read it and understand the business in five seconds. A machine reads the same page and has to guess.

Structured data is how you stop making it guess.

What Structured Data Actually Is

When a crawler hits your website, it sees text. Lots of text. It has to figure out, on its own, which words are your business name, which string is a phone number, whether “Tuesday 9 to 5” means your hours or a description of a service, and whether those five-star icons are real reviews or decoration.

Sometimes it gets that right. Often it doesn’t.

Structured data removes the guessing. It’s a small block of machine-readable code — usually a format called JSON-LD, based on a shared vocabulary called schema.org — that sits invisibly in your page’s source. Your visitors never see it. But every crawler and AI assistant reads it first, because it’s labeled.

Instead of hoping a machine infers your phone number, you hand it over, tagged: this is the phone number. Instead of hoping it understands you’re a plumber in a specific county, you state it: this is a LocalBusiness, here’s the name, here’s the area served, here are the hours.

Human-readable says “we’re open Tuesday.” Machine-readable says "openingHours": "Tu 09:00-17:00". One is a sentence to interpret. The other is a fact to file.

It’s the difference between leaving a note on the counter and filling out the form.

The Four Schemas That Matter for a Local Business

You don’t need to know how to write any of this — that’s our job. But it helps to know what’s being described, because these four cover almost everything a service business needs.

LocalBusiness

The foundation. This tells systems what you are and where you operate: business name, address, phone, service area, hours, the category you fall into. This is what makes you eligible to be the answer to “near me” searches. Without it, you’re a pile of text a crawler has to reverse-engineer into a business.

Service

This breaks down what you actually do. Drain cleaning, water heater installs, emergency calls — each one tagged as a distinct service. When someone asks an assistant for a very specific need, the systems matching that request to a business can see your services spelled out instead of buried in a paragraph.

FAQ

Your real answers to real questions — service area, whether you do emergency work, how booking works, what to expect on a first visit — marked up so a machine can lift them directly. This is increasingly how AI assistants build their answers. When ChatGPT or Perplexity Perplexity responds to “do plumbers in this area do weekend calls,” it’s pulling from sources that stated it plainly. FAQ schema is you stating it plainly, in a format built to be quoted.

Review

Your reputation, made legible. Star ratings and testimonials tagged so they register as actual trust signals instead of stray numbers on a page. This is what can put rating stars next to your listing in search results — and it’s a signal the recommendation systems weigh when deciding who’s worth surfacing.

Here’s roughly what a slice of LocalBusiness markup looks like under the hood:

{
  "@context": "https://schema.org",
  "@type": "Plumber",
  "name": "Example Plumbing Co.",
  "areaServed": "Example County",
  "telephone": "+1-555-0100",
  "openingHours": "Mo-Fr 08:00-17:00",
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.9",
    "reviewCount": "120"
  }
}

You never see it. The machine sees nothing but it.

Why Almost Nobody Has It

Two reasons.

First, it’s invisible. There’s no payoff you can look at. A new logo, a redesigned homepage — those feel like progress. Structured data has no visible surface, so it never makes the to-do list. Nobody brags about their JSON-LD.

Second, most website builders don’t add it for you. Drag-and-drop platforms hand you a pretty page and stop there. The machine-readable layer — the part that decides whether crawlers and assistants actually understand your business — gets left out, because the people building those templates were optimizing for what looks good in a demo.

So you end up with the common situation: a perfectly nice-looking site that is, to every system deciding who gets recommended, ambiguous. It exists, but it doesn’t explain itself. And ambiguous loses to specific every time a machine has to pick.

This Is How You Become Recommendable

Search ranking used to be the whole game. It still matters — but it’s no longer the only door.

People now ask an assistant: “who’s a good electrician near me?” “find me a contractor open right now.” The assistant doesn’t show ten blue links. It picks. And it picks from sources it can read cleanly and trust — the ones that stated their name, location, services, hours, and reputation in a format built for machines.

If your competitor’s site says all of that in structured data and yours leaves it to interpretation, the assistant has an easy choice and a hard one. It takes the easy one.

This is the quiet shift. Being human-readable gets you a site people can use once they find you. Being machine-readable is what determines whether you get found — and now, whether you get recommended out loud by the AI tools more of your customers reach for first.

The good news: it’s a one-time foundation, mostly invisible, and it works around the clock for every crawler and assistant that comes knocking. The bad news is only that almost nobody bothers — which is also the opportunity, because the bar to stand out is lower than it should be.

Get the Layer Machines Read

A good website has two audiences. The one you can see — your customers — and the one you can’t, the systems that decide whether those customers ever find you. Most sites are built for the first and forget the second entirely.

We build for both. Every site we ship includes the structured-data layer that tells Google Google and the AI assistants exactly what your business is, where it serves, and why it’s worth recommending — so you’re not a pile of text a machine has to guess at.

See how we build websites that machines can actually read →