If you ran a Google search in 2023, you saw ten blue links. If you run the same search today, you see an AI-generated paragraph at the top that cites two or three sources, pushes organic results below the fold, and answers the question without requiring a click. That AI paragraph is fed, in large part, by structured data — schema markup. The businesses being cited are not always the most authoritative. They are often simply the ones whose pages are correctly tagged.
Schema markup is not a new concept, but its strategic importance has changed dramatically. In the era of ten blue links, schema helped your listing look richer in search results. In the AI Overview era, schema is one of the primary signals that determines whether an AI model extracts your information, attributes it to you, and presents it as an answer. This article walks you through the exact implementation every local and regional business needs in 2026.
How AI Search Engines Use Structured Data
Large language models are trained on text. When they are deployed as search assistants — Google Gemini, Bing Copilot, Perplexity — they retrieve pages and extract facts to build answers. Plain text requires the model to infer what a piece of information represents. Schema markup removes the inference step. When your page says "priceRange": "$100-$500" in a LocalBusiness schema block, the AI does not have to guess your prices from a paragraph. The value is labeled, machine-readable, and extractable.
The practical result: pages with correct schema markup are cited in AI Overviews and conversational search answers at significantly higher rates than unstructured equivalents. Google's documentation confirms that structured data is one of the signals its systems use for AI Overview generation. This is not a hypothesis — it is documented behavior.
The Five Schema Types Every Local Business Needs
1. LocalBusiness (or subtype). This is the foundation. It tells AI models your business name, address, phone number, hours, service area, and category. Use the most specific subtype available: Plumber instead of LocalBusiness, AutoRepair instead of AutomotiveBusiness. More specific types carry stronger signals.
LocalBusiness schema — minimum viable implementation:
{
"@context": "https://schema.org",
"@type": "Plumber",
"name": "Your Business Name",
"address": {
"@type": "PostalAddress",
"streetAddress": "1234 Main St",
"addressLocality": "Houston",
"addressRegion": "TX",
"postalCode": "77001",
"addressCountry": "US"
},
"telephone": "+17135550100",
"openingHours": ["Mo-Fr 08:00-18:00", "Sa 09:00-14:00"],
"priceRange": "$$",
"url": "https://yourbusiness.com"
}2. FAQPage. This is the single highest-value schema type for AI Overviews. Each question-answer pair is a candidate for extraction. When a user asks a question your FAQ covers, your FAQ answer is a structured, labeled, attributable response the AI can use. The key rule: the visible <h5> or <h3> question text on your page must match the name field in your JSON-LD exactly.
3. HowTo. If any of your services can be explained as a process — how to get a quote, how your installation works, how to prepare your home for a service call — wrap that content in HowTo schema. Each step becomes a labeled, extractable unit. AI models use HowTo schema heavily when answering process questions.
4. Review / AggregateRating. If you have reviews, structure them. An AggregateRating block tells AI models your average rating and review count without requiring them to scrape a review platform. It appears in rich results and strengthens your credibility signal in AI-generated summaries.
5. BreadcrumbList. Breadcrumb schema tells AI models where a page fits in your site hierarchy. It enables rich breadcrumb display in search results and provides site structure context that helps models understand your content's topical authority.
Implementation: JSON-LD in the Page Head
Always use JSON-LD format, not microdata or RDFa. Google recommends JSON-LD and it is the format AI models and modern search systems are optimized for. Place all schema blocks inside <script type="application/ld+json"> tags in the <head> section of your page.
For a service page, implement three schema blocks in this order: Article or Service schema first, BreadcrumbList second, FAQPage third. This order does not affect parsing but helps your development team maintain consistency.
FAQPage Schema: The Highest-ROI Implementation
Most business websites have zero FAQ pages. Of those that do, fewer than 5% have FAQPage schema. This gap is your opportunity. A well-written FAQ page with correct schema implementation creates multiple AI Overview citation candidates at minimal cost. Here is the pattern:
- Identify the 5-8 questions your customers ask most often before buying.
- Write clear, factual answers of 40-80 words each. Avoid marketing language.
- Add visible
<h5>headings for each question on the page — exact text matching the JSON-LDnamefield. - Add a FAQPage JSON-LD block in the head with all questions and answers.
- Validate with Google's Rich Results Test.
The questions that generate AI Overview citations are not always the ones you would expect. "How much does [service] cost in [city]?" and "How long does [service] take?" are extremely high-frequency and almost never answered with structured data. If you answer those questions with FAQPage schema, you are competing for AI citations that your competitors are simply not in the running for.
HowTo Schema for Service Businesses
A plumber, HVAC company, or roofer that explains their process in HowTo schema is answering queries like "how does a plumber fix a burst pipe" and "what happens during an HVAC tune-up" — queries their potential customers are asking AI assistants. Each step in your HowTo schema is an extractable unit. If your process has 5 steps and your competitor has no HowTo schema, your 5 steps are the answer and your name is the attribution.
The HowTo schema does not require a dedicated "Our Process" page. It can be embedded on any service page. A roofing company can add a HowTo block to their roof replacement page explaining the inspection, removal, installation, and cleanup steps. That structured content feeds AI answers without requiring a separate content investment.
Common Schema Errors That Kill Your AI Visibility
Mismatch between visible text and JSON-LD. If the question visible on your page says "How much does a roof replacement cost?" but your JSON-LD says "What is the cost of a roof replacement?", both may be semantically similar but technically different. Search systems prefer exact matches. Use the same string in both places.
Using microdata instead of JSON-LD. Microdata is embedded in HTML attributes. JSON-LD is a separate script block. Microdata is harder to maintain, more error-prone, and not recommended by Google for new implementations. If your site uses microdata, migrate to JSON-LD.
Missing telephone format. Phone numbers in LocalBusiness schema should use E.164 format: +12125551234 for US, +5255123456 for Mexico. Do not use parentheses, dashes, or spaces. AI extraction systems parse phone numbers from structured data and display them directly in search results.
Stale opening hours. If your LocalBusiness schema says you are open Monday to Friday 9-5 but you actually close at 4 on Fridays, AI models will present incorrect information attributed to you. Update your schema whenever your hours change.
Schema for Multi-Location Businesses
If you operate in multiple cities — or serve a regional market across Houston, Dallas, San Antonio, or across Monterrey, Guadalajara, and CDMX — you need location-specific schema on location-specific pages. A single LocalBusiness block on your homepage does not adequately signal geographic coverage to AI models.
Create a dedicated page for each primary service area. Each page gets its own LocalBusiness schema with the specific address, phone, and service details for that location. This creates a structured data signal at the page level that AI models can associate with city-specific queries.
Validating and Monitoring Your Schema
After implementation, validate every page with Google's Rich Results Test at search.google.com/test/rich-results. Run the test by URL, not by code paste — the URL test catches rendering issues that a paste test misses.
Check Google Search Console under Enhancements for structured data reports. GSC shows which schema types are detected across your site, how many pages have errors, and whether those errors are causing rich result disqualification. If you have FAQPage schema on 10 pages and GSC shows 3 errors, those 3 pages are not generating FAQ rich results.
Schema is not a one-time setup. When you add pages, change services, update prices, or adjust hours, your schema needs to reflect those changes. Build schema updates into your content workflow the same way you build in meta description updates.
The Competitive Moat Schema Creates
The gap between businesses that implement schema correctly and those that do not is widening in 2026. AI search is not going away — it is becoming the primary interface. The businesses cited in AI Overviews get the query resolution without the click. The businesses not cited compete for whatever organic traffic is left below the fold.
Schema implementation requires a few hours of development work and a basic understanding of JSON-LD syntax. Once implemented correctly, it compounds: each new page with proper schema is another AI citation candidate. A business with 20 well-structured service pages and FAQ sections has 20 structured chances to appear in AI-generated answers for their most valuable queries.
Your competitors are not doing this. The Rich Results Test will confirm it in minutes — run it on five competitor URLs and see how many have zero structured data. That gap is your opportunity window, and it will not stay open forever.
What is schema markup and why does it matter for AI search?
Schema markup is structured data in JSON-LD format that tells search engines and AI models exactly what your content means. AI search engines like Google Gemini and ChatGPT read schema to extract facts about businesses, services, prices, and reviews without having to guess from plain text.
Which schema types are most important for local businesses in 2026?
LocalBusiness or its subtypes (AutoRepair, Restaurant, Plumber, etc.), FAQPage, HowTo, Review, and BreadcrumbList are the five most impactful. LocalBusiness provides NAP data. FAQPage and HowTo feed directly into AI answer extraction.
Does schema markup help with Google AI Overviews?
Yes. Google's own documentation confirms that structured data helps AI Overviews extract and attribute information. Sites with FAQPage and HowTo schema are significantly more likely to be cited in AI-generated summaries than unstructured pages.
How do I test if my schema markup is valid?
Use Google's Rich Results Test at search.google.com/test/rich-results. Paste your URL or code snippet and it shows which schema types are detected and any errors. Schema.org also has a validator at validator.schema.org.
Can I use schema markup on a website I did not build myself?
Yes. Add a JSON-LD script block to the HTML head of any page you control. WordPress users can use the Rank Math or Yoast SEO plugin to add schema without touching code. Shopify users can add it via theme liquid files or a custom script app.