SEO and AI search in 2026: what actually has to be on a home service site

Search for a plumber, a roofer, or an HVAC shop is no longer only a list of blue links. A homeowner still Googles the job and the town. A growing share also asks ChatGPT, Gemini, or the AI overview sitting on top of the results. If your shop is missing from either place, you are not in the conversation.

That is not two products. Google and the assistants read the same site. They punish the same gaps: a slow page, a name that does not match the Google profile, a service page that never says the town, a robots file that blocks the crawler.

For an owner, this is not a marketing-theory problem. It decides whether Thursday's board has work on it. Rankings and impressions are not enough in a world where the first thing a homeowner sees is a synthesized answer, or three names in the map pack, and they call from there.

Shops that make the business readable, specific, and consistent, then keep publishing proof they do the work in that town, show up in both places. The ones that buy a keyword package and leave the profile empty do not. The rest of this article is what that work actually is.

The goal is to be the answer, not only the first link

The map pack and the organic listing still send calls. They are not the whole path anymore. An AI overview can name two companies and a phone number before anyone clicks. An assistant can recommend "a plumber in St Charles" from whatever facts it could extract last time it looked.

If those facts are missing, conflicting, or written as slogans, the model hedges or picks the shop whose pages are unambiguous. You do not have to rank first to lose the job. You only have to be unreadable.

What SEO and AI search optimization means for a contractor

For a home service company, SEO is the work of showing up when someone in your town types the job they need. AI search optimization is the work of making the same facts extractable: who you are, what you do, where you do it, the number to call, and whether real people have hired you.

Traditional SEO asked whether a URL ranked. This still matters. The extra question is whether a system that never shows the full page can quote you correctly. That is a content and identity problem, not a keyword-density problem.

Three things the page has to get right

You can ignore most of the acronyms. Three properties decide whether Google and an assistant can use you.

• Extractability. Facts in plain sentences and in structured markup: business name, phone, towns served, job types. Headings that name the question. A section that can stand on its own if it is lifted out of the page.
• Verifiability. The same legal name, the same number, and the same towns on the site, in the markup, and on the Google Business Profile. Reviews in volume. Real photos of jobs you finished. "Apex Caulking" on the page and "Apex Caulking LLC" on the profile is the mismatch that makes a model refuse to recommend you.
• Contextual clarity. The first line says the trade and the town. Drain cleaning is not water heater installation. Each service page answers the questions a homeowner actually asks, with the place named, not a 200-word stub with the city stuffed in three times.

How this is different from "doing SEO" as a monthly retainer

Old local SEO was often a citation blast, a plugin, and a report of rankings. AI-aware search still needs those directories to agree. It also needs the site itself to be a set of facts a machine can trust.

Assistants do not "like" your brand. They assemble answers from pieces. A buzzword on a services page ("fully optimized for AI search") is not a fact they can use. A LocalBusiness block that matches the Google profile is. A heading that is a real question, with a short honest answer underneath, is. A robots file that does not block the crawler is.

Skip the tricks that do not work. Publishing an llms.txt file is not a strategy. Google's search team has said they do not use it, and large crawls have found that almost none of those files are even requested. Identity is the work.

Why this matters before you spend a dollar on ads

A homeowner with a real problem does not tour ten websites. They look at the map, the stars, and whatever the overview said. If you are not in that set, paid search is how you rent a place in it, at a premium, every week.

Local search is the free inventory. Service pages, the Google profile, reviews, and markup keep paying after you stop writing checks. Ads can put you in front of people this week. They are expensive waste if the profile is empty and the site cannot be parsed.

There is a second cost. If an assistant invents your hours, your towns, or your phone number, the customer may form a decision you never get to correct. Clear, matching facts are how you stop that. You cannot buy a citation in ChatGPT. You can make the shop unambiguous enough that when someone in your town asks for a roofer, the model has something true to say.

Practices that still work if you never hire anyone

Do these in order. They help Google first, and they are what the assistants read second.

Finish the Google Business Profile. Categories for the jobs you want, hours that are true including holidays, photos from the last 90 days, a description that names the town and the work, posts that actually go out.

One page per job you want to win, with the town on it. At least a few hundred words of useful content, a visible number, and questions a tech already hears, written as questions, with answers underneath. Two such pairs on a service page is a start. Fake FAQ markup for a rich result Google no longer shows is not the point. The sentences on the page are.

Put LocalBusiness markup on the site with the real name, the real phone, and the towns you actually serve. Match the profile. Do not smooth away LLC. Do not invent a town.

Make name, address, and phone identical on the site, the profile, Yelp, Angi, and the BBB. "St." versus "Street" is enough to suppress you.

Write for the search they type before they know your name. "Why is my furnace not producing heat" is a different query from your company name. Two real posts a month is the floor if you are doing it yourself. Use your photos. Link back to the service page that books the work.

Check robots.txt so you are not accidentally excluding the crawlers. Being blocked is invisible from the rest of the site, and it is how a shop disappears from Google and the assistants at once.

Make the site fast on a phone. More than half of mobile visits leave if the page takes longer than three seconds. A slow site does not rank its way out of that, and a model that never finishes loading you will not quote you.

What rankings will not tell you

Position for "plumber near me" is a useful number. It does not tell you whether the overview cited you, whether the map pack included you, or whether the name in the markup matches the name on the door. Impressions can rise while the facts a model would extract are still wrong.

Look at three things alongside rank: whether the profile is complete and active, whether each money job has a real page, and whether a stranger (or a scorecard) can tell the trade, the town, and the number in a few seconds. Those are the inputs. Rank is an output, and it lags.

How GoodLeed builds both

The Growth Engine treats SEO and AI search as the Found job: show up when a homeowner needs you. One team, one site, one set of facts. Not a website guy, an SEO vendor, and a third invoice for AI.

Every client site is hand-coded. PageSpeed on goodleed.com is the build standard those sites get. 100/100 on Google's SEO and best-practices audit is the floor we quote because you can run it yourself, not because it looks good on a slide.

Schema and structured content are built in: LocalBusiness markup with the real name, number, and towns; service pages and city pages as actual pages; questions a tech already hears, on the page. We verify the markup against the facts we hold for the business, and we keep checking after launch, because a redeploy can silently take it out.

Google Business is on the same calendar as the rest of the content: three profile posts a week, plus three blog posts and three social posts. Nine or more pieces every week. Ads wait until this layer converts, in your own ad account, never marked up.

The first 30 days are that foundation, plus the review engine and the 60-second response. Rankings and reviews compound after that. You watch it on a live dashboard, not a monthly PDF of impressions.

SEO and AI search FAQs

Do I still need regular local SEO if people are asking ChatGPT?

Yes. Assistants draw from the same site, the same profile, and the same reviews Google uses. A finished Google Business Profile, matching name and phone, and real service pages are still the inventory. AI search does not replace that work. It punishes you faster when that work is sloppy.

Will schema alone get me into AI answers?

No. Markup helps a system find the facts. The facts still have to be true, matching, and written on the page in language a person would use. Schema pointing at a placeholder name or a town you do not serve is worse than none.

Should I publish an llms.txt file?

Not as a strategy. Google has said it does not use the file. Studies of large numbers of domains have found that almost none of those files are requested. Spend the time on matching identity, service pages, and a robots file that does not block crawlers.

When should I turn on Google ads?

After the organic path converts: fast site, finished profile, real service pages, and an answer when they call. Paid search can be the highest-return channel in the trades. It is expensive waste if you are buying clicks into a shop nobody can parse.

How do I see whether my shop is readable today?

Run the Local Growth Scorecard. It grades Found, Chosen, and Booked in about a minute. Found is the half this article is about. Booked is still the half that turns a ranking into a job on the calendar.