What Should Location Pages Include for AI Search? 2026 Data

A location page for AI search should be a page about one service in one city, carrying that branch's facts, proof and a way to book, because that is the kind of page AI engines cite. Of 74,969 citations that ChatGPT, Gemini, Perplexity and Google AI Mode gave to contractor websites while answering 32 home-services questions in 100 US and Canadian metros between August 31 and September 20, 2026, 43% went to homepages and 48% to service, location or service-plus-city pages. Blog posts got 1.5% and review or testimonial pages 0.7%.

For a brand with 40 branches, that makes the page set a practical question. A single locations page that lists addresses gives the engines nothing to cite when a customer asks for the best water heater repair in one of your cities, and a blog post about water heaters barely registers.

Last verified: September 23, 2026.

The numbers come from the Cheers Market Baseline, which asks four AI engines the same questions every week; the method is at the end of this page.

Which pages AI engines cite on contractor websites

We sorted every cited contractor URL by its path. The homepage took 43.1% of the citations. Pages that named a service and a city together, such as a water heater page inside a city folder, took 17.1%. Service pages with no city took 16.9%, and location or city pages with no service keyword took 14.0%. Everything a marketing team tends to produce for "content" came last: booking, offer and financing pages 2.3%, blog and resource pages 1.5%, about and contact pages 1.2%, and review or testimonial pages 0.7%, with 3.2% left unclassified.

The trade shifts the balance but not the order. In plumbing, 35.7% of the 25,033 cited URLs were homepages, 19.6% service pages, 18.3% service-plus-city pages and 16.4% location pages. In HVAC it was 45.8%, 17.3%, 15.4% and 10.5%, and in electrical 48.8%, 24.7%, 13.4% and 6.6%. Roofing, with two questions and 2,631 cited URLs, leaned on homepages at 67.7%.

Each engine reads a different page

The blended numbers hide a split between Google AI Mode and everyone else.

Perplexity cited service-plus-city pages more than any other type, 28.6% of its 23,011 website citations, then service pages at 26.5%, homepages at 19.6% and location pages at 16.5%. Gemini was close behind: homepages 29.6%, service-plus-city pages 26.0%, service pages 23.1% and location pages 13.5% of 9,341 citations. ChatGPT sat in the middle, with homepages at 37.4%, service pages 24.1%, service-plus-city pages 17.4% and location pages 10.7% of 11,752 citations.

Google AI Mode sent 66.9% of its 30,865 website citations to homepages, 13.5% to location pages, 5.7% to service-plus-city pages and 5.0% to service pages. Our read, which is an inference and not something the data proves, is that most of those homepage links come from the website button on each business's Google Business Profile. If your profiles all point at the brand homepage, Google AI Mode has no branch page to send people to. Where each profile should link is covered in linking a Google Business Profile to its location page, and Google's business links policy already requires it for booking and ordering links:

For businesses with multiple locations, action links must lead to a website for a specific location.

The page spec for a multi-location brand

This is the page we would build for each service a branch actually runs, in the city it serves, and the evidence behind each part.

Name the service and the city

The URL, title and H1 should say the job and the place, for example /locations/denver/water-heater-repair. That is the page type Perplexity cited most (28.6% of its website citations) and Gemini second most (26.0%). One page per service the branch really performs is enough; a page for every suburb is not the goal.

Carry the branch facts, word for word

Branch name, address or a plain description of the service area, the local phone number, hours and the towns the crew covers. These should match the branch's Google Business Profile exactly, because the profile and the page are two copies of the same record, and an engine that finds two versions has to pick one.

Answer the questions the engines answer

Say what this branch does for this service: the scope, the brands it services, whether it takes emergency calls and when, how pricing works and what the warranty covers. The answers themselves talk about these things. For emergency questions, 93% to 99% of answers on every engine mentioned 24/7 or after-hours availability, against 34% to 61% for "best company for this job" questions, and for deal and quote questions ChatGPT and Gemini put a dollar figure in 62% and 68% of answers. Those are counts of what the answers mention, not proof of why a business was picked, so publish the facts that are true for the branch and leave out the ones that aren't.

Put the proof on the page

Recent reviews that mention this service in this city, photos of real jobs, the technicians who do the work and current licenses belong on the service-plus-city page itself. A standalone testimonials page took 0.7% of website citations and blog posts 1.5%, so proof parked there is proof the engines rarely read.

Give it one way to book

Link the same online booking page the branch's profile uses, and show the local phone number. Google AI Mode linked a business's online booking page in 31.2% of its answers, most often on Housecall Pro (20.5% of answers), ServiceTitan (10.6%) or Jobber (3.1%). That these links come from the Book button on each Business Profile is our inference. ChatGPT and Gemini never linked a booking page in this sample, and Perplexity did in 0.2% of answers.

The trade matters. Across plumbing's 9 questions, 39.4% of Google AI Mode answers carried a booking link, HVAC's 9 questions 38.9% and electrical's 6 questions 33.3%. Roofing, with 2 questions, was at 1.8%. If your trade books online, the booking page is part of the location page's job.

Mark up what the page shows

Add LocalBusiness structured data for the branch with the same name, address, phone and hours the page displays. Google's AI features guide is explicit that nothing extra is needed for its AI surfaces: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them." The markup describes the page; it doesn't replace it. For multi-location markup patterns, see LocalBusiness schema for multiple locations.

What not to build

A page for every town, with the city name swapped and nothing else changed, is the pattern Google's spam policies call doorway abuse. The policy, last updated August 28, 2026, lists pages "targeted at specific regions or cities that funnel users to one page" and "substantially similar pages that are closer to search results than a clearly defined, browseable hierarchy." Our data shows engines citing pages that name a service and a city; it does not show them rewarding a hundred near-identical copies. Strip out the local facts and proof and you have the doorway version. Building city pages without doorway risk walks through where that line sits.

What to do this week

Pick one market and one high-value service, and get that single page to spec before you touch the rest.

• Export every Business Profile's website link and booking link, and mark which ones point at the brand homepage instead of the branch page.
• List the services each branch actually performs and check that each has its own page with the city in the URL, title and H1.
• For the chosen market, rewrite the page to the spec above: branch facts matching the profile, service specifics, reviews and photos from that city, one booking link, and matching LocalBusiness markup.
• Ask the four engines your money question for that city before and after the change, and save which of your pages they cite. The free AI Visibility Grader runs a one-location check on ChatGPT, Gemini and Perplexity.

If you run dozens of markets and would rather have one team keep every branch page and profile in line, that is what we do in multi-location local SEO.

Methodology

Every week, Cheers asks ChatGPT, Gemini, Perplexity and Google AI Mode the same 32 home-services questions in 100 metro areas across the United States and Canada, and records every business each engine names, with its star rating, review count and map position where the engine shows one, and every source it cites. This analysis covers the three weeks from August 31 to September 20, 2026: 34,742 usable answers (9,434 from ChatGPT, 9,435 from Google AI Mode, 9,364 from Gemini and 6,509 from Perplexity) across 60 US and 40 Canadian metros. We excluded empty or cut-off answers, most of them from a Perplexity collection outage. Each question was asked once per city per week, so week-to-week changes include normal answer-to-answer variation. Business names were extracted automatically; a hand check of 40 answers found the extraction caught about three in four of the businesses a reader would count, so counts are lower bounds. Star ratings and review counts are the figures each engine displayed; for ChatGPT and Gemini we captured them from September 18 and September 14 onward. Findings describe associations in this sample, not how any engine ranks businesses.

Page types come from URL paths, first match wins. A homepage has an empty path, /, /index or /home. Review pages have "review" or "testimonial" in the path, and blog pages sit under paths such as /blog, /news, /resources or a dated folder. A service-plus-city page has a location folder such as /locations/ or /service-areas/, or the answer's city in the path, plus a trade or service keyword; a location page has the folder or city without a service keyword; a service page has the keyword alone. Citations were counted once per answer and only on domains that match a business named somewhere in the dataset. Full definitions are on our research methodology page.

Sources

Every link below was opened and checked on September 23, 2026.

• Business links policies & guidelines. Google Business Profile Help. The rule that multi-location action links must lead to a page for a specific location, and that links must let customers complete the action.
• Spam policies for Google web search. Google Search Central, last updated August 28, 2026. The doorway abuse examples quoted above.
• Optimizing your website for generative AI features on Google Search. Google Search Central, last updated July 10, 2026. No special files or markup for Google's AI features; Business Profile data feeds them.
• Local business (LocalBusiness) structured data. Google Search Central, last updated September 8, 2026. The LocalBusiness markup reference.
• Cheers Market Baseline, August 31 to September 20, 2026: 32 home-services questions, 100 US and Canadian metros, 34,742 usable answers, 74,969 citations to contractor websites, aggregate only. Research methodology.

Dylan Allen-Arnegård is the CEO & Co-Founder of Cheers, the local search platform for multi-location service businesses.