The shift nobody's talking about

For 20 years, local search worked the same way. You optimize for Google. You aim for the local 3-pack or position #1 in organic results. Customers click. Some convert.

Position #1 in Google Maps captures roughly 100% of the click traffic in its category. Position #2 captures about 32%. Position #8 captures 11%. Position #20 captures 1%.

Those numbers haven't changed in years. What HAS changed is that an increasing share of customers no longer search Google at all. They open ChatGPT, Perplexity, or Google's AI Overviews and ask:

"Who's a good plumber in Hartford?"

"What's the best vet near me for an older dog?"

"Recommend an auto repair shop that's good at older Volvos."

The AI doesn't return ten links. It returns a recommendation — usually 1-3 businesses by name, with a sentence about each.

The conversion math is wildly different

When a customer clicks your Google listing, they're entering your funnel — landing page, contact form, phone call, conversion. There's friction at every step. The standard local-services conversion rate from organic Google click to actual customer is 3-8%.

When ChatGPT says "I'd recommend Acme Plumbing — they have great reviews and are highly rated for emergency calls" — that's not the start of a funnel. That's a direct endorsement from a tool the customer already trusts.

Customers who hear a specific recommendation from an AI engine convert at a much higher rate, in our analysis-based modeling: somewhere in the 18-30% range, depending on category. That's because:

  1. The friction is gone. They're not comparing ten options; they have an answer.
  2. Trust is transferred. ChatGPT recommended you specifically; the customer treats that as social proof.
  3. The competitor scan is shortcut. They don't see Bob's Plumbing right next to yours in a list.

A ChatGPT recommendation isn't traffic. It's intent that's already past the comparison stage.

What "being recommended by name" actually requires

AI engines build their recommendations from a structured understanding of your business. The signals that matter most:

1. Niche-specific schema markup

Generic LocalBusiness schema is the floor. The lift comes from using Schema.org's specific subtypes:

  • A vet uses VeterinaryCare, not LocalBusiness
  • A funeral home uses FuneralHome
  • An auto repair shop uses AutoRepair
  • An accountant uses AccountingService
  • A pet groomer uses PetStore (or the more specific AnimalShelter if applicable)

When an AI engine sees VeterinaryCare, it knows EXACTLY what to recommend the business for. When it sees generic LocalBusiness, it has to guess.

Most local business websites use generic markup. A Wix or Squarespace template gives you LocalBusiness and stops there.

2. Entity linking with @id

If your homepage describes your business, your contact page describes the same business, and your services page lists what that business offers — AI engines need to know it's all the same entity.

The @id field in JSON-LD links these together. Without it, the AI sees three separate "businesses" and may discount you as low-confidence.

3. Active review data

AI engines weight aggregate review count and rating. They also weight recency. A business with 287 reviews and the most recent posted three days ago looks much better than 312 reviews with the most recent posted seven months ago.

4. Plain-language FAQ content

ChatGPT, Claude, and Google AI Overviews extract answers from FAQ schema and from H2/H3 question-answer pairs. The questions need to match how customers actually phrase queries:

  • ✗ "Pricing" (you optimized for this for Google)
  • ✓ "How much does it cost?" (this is what AI engines extract)

5. llms.txt at site root

A plain-text file summarizing your business, services, and key facts. AI crawlers read it preferentially when building their index.

Most local business sites don't have this. The few that do get cited disproportionately.

The differentiator window

This is the structural opportunity for local businesses right now:

Most of your competitors aren't doing any of this. They have generic schema (or none). They don't have an llms.txt. Their FAQ is keyword-stuffed Google bait, not plain-language Q&A. They've never set up entity linking.

That means the local business that DOES this work in 2026 gets cited by name when customers ask AI engines for recommendations — while their competitors don't even appear.

The window closes in 12-24 months minimum. Once Wix, Squarespace, and WordPress bake this into their default templates, the structural advantage normalizes. The businesses that established their AI-search presence early will have years of citation history baked in. The ones that wait will be playing catch-up against entities the AI already knows about.

The gut check

If you're a local service business and you've never had your site audited for:

  • The specific Schema.org subtype for your niche
  • @id entity linking across pages
  • An llms.txt file
  • AI-crawler-friendly robots.txt
  • A FAQPage schema block tied to actual customer questions

…then customers asking ChatGPT for a recommendation in your category right now are getting your competitors' names, not yours.

That's not a problem you fix with more Google ads. That's a problem you fix by getting your structured data right.

If that's where you are, we'll show you exactly what's missing for free. No credit card. No contract. We earn the work — or we don't keep you.