Schema Markup Without the Jargon (and Why It Matters More Than Ever for AI)

An SEO consultant told a buddy of mine his plumbing site needed "schema markup." He paid the audit invoice and three months later still had no idea what he'd actually bought. He asked me one night at a job site: is schema a thing I should care about, or is it a term agencies use to make a $4,000 retainer sound like science? Both, sort of. The concept is genuinely useful. Almost nobody bothers to explain it in plain English to the owner who has to fund it.

Schema is having a quiet moment right now because of AI search. Google's AI Overviews, ChatGPT Search, Perplexity, and Gemini are all leaning on structured data to figure out what your business is and whether to cite you. The difference between getting cited and getting skipped is often whether your site speaks the machines' language. So this post is the conversation I had with my buddy that night, written down. No jargon you can't define out loud. One real example. And a checklist you can run on your own site this month.

What is schema markup, in plain English?

Schema markup is a set of invisible labels you put on your website so search engines and AI systems know what each piece of text actually means. The price on your service page is labeled "price." Your address is labeled "address." Your phone number is labeled "telephone." Without those labels, the text is just a string of characters a machine has to guess about. With them, the machine knows for certain.

Search engines used to read the web the way you read a book, scanning text and inferring meaning. That works for humans. It works less well for software trying to answer "what's the phone for Joe's Plumbing in Sandy?" The software has to find a phone-number-shaped string and hope it picked the right one. Schema removes the guessing. You wrap the number in a label that says "this is the telephone for this business," and the machine reads it directly.

The technical name for this label set is structured data. Schema.org is the shared vocabulary Google, Microsoft, Yahoo, and Yandex agreed on in 2011. When SEO people say "schema," they mean schema.org structured data. The terms are interchangeable.

How does schema actually live on my page?

There are three formats for schema. Two are old, one is current. The current one is JSON-LD, which stands for JavaScript Object Notation for Linked Data. It's a small block of structured text that lives in your page's HTML, usually inside the head section, invisible to anyone reading the page in a browser. Search engines and AI crawlers read it directly and store the result as facts about your business.

The two older formats are microdata and RDFa. Both stuck labels into your visible HTML and both are basically dead in 2025. Google's official structured data documentation recommends JSON-LD as the preferred format. If anyone proposes adding microdata or RDFa to your site in 2026, push back.

Here's what a real JSON-LD block looks like for a fictional plumbing company:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Plumber",
  "name": "Joe's Plumbing",
  "image": "https://joesplumbing.com/logo.png",
  "telephone": "+1-801-555-0142",
  "priceRange": "$$",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "1234 Main St",
    "addressLocality": "Sandy",
    "addressRegion": "UT",
    "postalCode": "84070",
    "addressCountry": "US"
  },
  "geo": {
    "@type": "GeoCoordinates",
    "latitude": 40.5649,
    "longitude": -111.8389
  },
  "openingHours": "Mo-Sa 07:00-19:00",
  "areaServed": ["Sandy", "Draper", "Cottonwood Heights", "Midvale"],
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.8",
    "reviewCount": "127"
  }
}
</script>

Read it line by line. The "@context" line names the vocabulary (schema.org). The "@type" line declares what this thing is (a Plumber, a sub-type of LocalBusiness). Every line after is a labeled fact: Joe's name, phone, price tier, address, coordinates, hours, cities served, and aggregate review rating. That's schema. A structured list of facts a machine reads without guessing. Every search engine and AI system that crawls the page picks it up and stores those facts.

Which schema types does a home-services site actually need?

For a typical home-services site, five schema types do almost all the heavy lifting: LocalBusiness, Service, FAQPage, BreadcrumbList, and Person. A sixth, AggregateRating, is helpful when you have real reviews to display. You don't need fifty types. You need these, on the right pages.

• LocalBusiness. Homepage and contact page. Declares your business with name, address, phone, hours, service area, and price range. Use the most specific sub-type that fits: Plumber, Electrician, RoofingContractor, HVACBusiness, GeneralContractor, HousePainter, MovingCompany. If your trade has no type, fall back to LocalBusiness. This anchors you as a real entity in the eyes of Google and AI.
• Service. Each service page. Declares what the service is and who provides it. A drain-cleaning page gets a Service schema with name "Drain Cleaning," provider pointing to your LocalBusiness, and areaServed listing your cities. Helps you show up for "drain cleaning Sandy" because the page is now explicitly tagged as offering that service in that area.
• FAQPage. Any page with a real Q&A section. Each question is labeled Question, each answer Answer. AI systems pull FAQ content into citations heavily. Note: Google narrowed FAQPage rich results in August 2023 so they only show for government and authoritative health sites. You won't see FAQ stars in your listing anymore, but AI crawlers still treat the schema as a strong citation signal.
• BreadcrumbList. Every interior page. Declares the page's hierarchy. Home > Services > Drain Cleaning. Helps search engines understand site structure and shows breadcrumbs in results instead of a raw URL.
• Person. About page and author bylines. Declares the owner as a real human with name, role, and worksFor pointing to your LocalBusiness. Ties your face to your business in the entity graph AI systems are building. Matters more every quarter.
• AggregateRating. Homepage or reviews page. Declares your star rating and review count. Be careful: Google requires the ratings be for the business itself, not stuffed in from third parties. If you have real Google reviews to point at, great. Don't fabricate. The penalty for fake review schema is harsh.

Six types, each tied to a specific page. A home-services site that ships all six is well ahead of the average for its category. Most local-business sites ship one or two schema types and stop, often whatever the site builder injects automatically.

Why does schema matter more for AI search than for regular Google?

AI Overviews, ChatGPT, Perplexity, and Gemini all use entity recognition to decide which businesses to mention in answers. Entity recognition means the system reads a page, decides what real-world things are mentioned on it (people, places, businesses), and builds a confidence score for each one. Schema is the highest-confidence signal those systems have.

Without schema, an AI has to infer that "Joe's Plumbing" is a business and not a song or product brand. It can usually figure that out, but the confidence is lower. Lower confidence means the AI is more likely to skip you when picking which business to cite. With schema, the inference goes away. Your JSON-LD says "@type: Plumber, name: Joe's Plumbing." That's a stated fact, not a guess. The AI stores it with high confidence and is far more likely to pull you into an answer for "who's a good plumber in Sandy?"

This matters more in 2026 than 2022 because AI summaries are eating the top of the results page. According to Search Engine Land's August 2025 reporting, Google AI Overviews now appear on roughly 31% of searches, up from 16% a year earlier. ChatGPT Search and Perplexity together pulled past 200 million weekly users in late 2025. The traffic that used to flow through ten blue links is increasingly flowing through AI answers, and the businesses cited are disproportionately the ones with clean structured data.

The other reason: AI systems don't store the web as a pile of pages. They store it as a knowledge graph, a network of entities (businesses, people, places) connected by labeled relationships (worksFor, areaServed, hasReview). Schema is the format that builds the graph. Sites without it show up as orphan nodes. Sites with full schema show up as well-connected entities. The graph is the substrate AI search runs on, and your schema is your placement in it.

How do I check what schema my current site has?

The fastest way is Google's Rich Results Test. It's free, official, and takes about 30 seconds. Open search.google.com/test/rich-results, paste your homepage URL, and click Test URL. You'll get a report listing every structured-data block on the page, what type each is, and any errors.

What you're looking for:

1. Detected items count. If zero, your page has no schema. The most common audit result for sites on Wix, Squarespace, or stock WordPress themes. You're an orphan in the entity graph.
2. Types found. If it only lists "Organization" or "WebSite," you have bare-minimum auto-generated schema. That's table stakes, not SEO. You want LocalBusiness (or a sub-type), FAQPage if applicable, BreadcrumbList, and Person.
3. Errors under each item. Errors mean Google discards the schema entirely. Fix immediately. Warnings mean it's incomplete but still valid. Address those when you have time.
4. Eligible for rich results. The tool tells you if your data qualifies for review stars, FAQ accordions, or breadcrumbs in search. Don't panic if FAQPage shows "not eligible." Google narrowed FAQ rich results in 2023, but the schema still helps for AI citations.

For a stricter read, run your URL through validator.schema.org. It flags issues Google's tool waves through, including type mismatches. For most home-services owners, the Rich Results Test is enough.

What does a properly-schema'd site look like in practice?

I rebuilt my own permanent-lighting site (TruLight SLC) on Next.js. The original site, built on Lovable.app, had basically no schema. The new build ships the full set: LocalBusiness with all sub-fields, Service schemas on each install-type page, FAQPage on the FAQ block, BreadcrumbList on every interior page, and Person for me as the owner with worksFor pointing to TruLight LLC.

The full breakdown lives at our TruLight SLC case study. The page-speed numbers are dramatic (load time dropped from 4,155 ms to 745 ms), but the schema work is what's still paying compounding dividends six months later. The site now shows up in AI Overview citations for "permanent lighting Salt Lake City" and adjacent queries the old site never touched. Same content, same offers. The difference is the AI systems can now read the page as structured facts.

The schema didn't take a separate engagement. Every Front Door Digital site ships with full structured data by default. There's no version of a modern web stack where adding schema is hard. The reason most local-business sites don't have it is that page-builder platforms either ship a thin auto-generated version or none at all, and there's no easy way for the owner to add it without writing code.

Frequently asked questions

Can I add schema to my Wix or Squarespace site without rebuilding?

Partially. Both platforms auto-generate a basic Organization schema and won't easily let you replace it. You can inject custom JSON-LD into the page header through advanced settings, but the auto-generated schema often conflicts with custom additions and creates errors. You can patch about 60% of what a custom-built site ships, and the patches are fragile. If schema is a real priority, the page builder is fighting you.

Will adding schema instantly help my Google rankings?

No. Schema is not a ranking factor like page speed or content quality. What it does is make your site eligible for rich results, improve AI entity recognition, and help search engines understand your business with higher confidence. Those effects compound over weeks and months. Don't expect to add schema on Tuesday and see a ranking jump on Friday. Expect AI citations and rich-result eligibility to improve over the next quarter.

Do I need to maintain my schema as my business changes?

Yes, but less than you'd think. The fields that change are hours, phone, address, and review count. On a properly-built site, those values come from a single source (your business profile or a CMS), so updating in one place updates everywhere including your schema. On a page-builder site, they're hand-typed into multiple places and drift apart over time. Either way: when your business changes, your schema should reflect it within a week.

Is schema markup worth paying an SEO consultant to add?

Depends what they're charging. Adding the basic six types to a typical home-services site is a few hours of work for a developer. If someone's quoting four figures for "schema implementation" alone, you're being marked up. A few hundred dollars as part of broader cleanup is reasonable. If you're already paying for a rebuild on a modern stack, schema should be included by default and you shouldn't pay extra.

If you got this far, the move this month is simple: open Google's Rich Results Test, paste your homepage URL, and see what's there. If the answer is "nothing" or "Organization only," you now know the exact gap your site has and why it matters. Fixing it on your current platform, hiring a developer, or rebuilding on something modern is a separate decision. The first step is looking at the receipts.