How a Restoration Company Reached #1 in AI Search in 21 Days

The short answer

A restoration company on Kauai went from zero AI visibility to the number one site ChatGPT, Perplexity, and Gemini cite for mold and water damage on the island, in 21 days, by rebuilding around three things: a site machine-readable enough to be quoted, structured data that spelled out every service and service area, and a citation trail across the sources those engines read. Twenty-one days after launch the company was cited ahead of ServPro, a national franchise with more than 2,400 locations, with mention rates of 77% on ChatGPT, 81% on Perplexity, and 83% on Gemini. This is a teardown of how it was built and what a home services owner can take from it.

Why we are writing this up

Most case studies in this industry show a number and hide the method. We would rather show the work, because the method is the point. AI search is where local discovery is moving. In BrightLocal's 2026 survey, the share of consumers using an AI assistant to find a local business jumped from 6% to 45% in a single year, which pushed AI past Yelp as a discovery channel. Restoration is one of the harder categories to win there, because the buyer is often mid-emergency and the national franchises are well funded. If the method holds up in restoration, it holds up in garage door, HVAC, plumbing, and the rest of home services.

What we started with

Kauai Mold Water Fire started at zero. When we ran the baseline, the AI engines did not name the company on a single one of the twelve real customer questions we track, from "best mold remediation company on Kauai" to "my house flooded on Kauai, who do I call." The engines were confidently naming other companies, plus the national franchise, plus a scattering of directories. The client did good work and had happy customers. None of that was legible to a machine yet.

What we built, and in what order

The build followed the same order we use on every home services engagement, because sequence matters. Structure before push.

1. A site built to be read. We rebuilt the site so every page was crawlable, indexable, and eligible to show with a snippet. Google is explicit that its AI features draw only from publicly accessible, crawlable content, so this is the floor. No shortcut skips it.
2. A direct answer on every page. Each service page opens with a short, self-contained answer to the exact question a customer asks, then expands underneath. AI engines lift standalone answers and skip buried ones.
3. Structured data for a local business. LocalBusiness, Service, FAQ, and Review schema told the engines the name, service area, hours, and the specific services, in the format they trust. Across the industry, verified structured data is the single most-cited source type, making up more than half of all citations in one study of 6.8 million of them.
4. A content architecture that answered the real questions. Not filler. Google's own guidance is to publish valuable, non-commodity content, so we built pages around the actual mold, water, and fire questions Kauai homeowners ask, with local specifics a generic page could never carry.
5. A citation trail. We made the business appear, described the same way, across the independent sources the engines read, so that every place the AI looked told the same story. Consistency is what turns a maybe into a recommendation.

The result at 21 days

Three weeks after launch, the picture had inverted. Kauai Mold Water Fire was the number one site the AI was learning from on the island, with 20 citations across the tracked prompts against ServPro's 15. On the twelve non-branded customer questions, the company was mentioned 77% of the time on ChatGPT, 81% on Perplexity, and 83% on Gemini, from a standing start of nothing. One local operator, on one island, was being cited more often than the biggest name in restoration.

The control test that proves it was the build

Kauai had off-page work layered on top, so a skeptic could argue the reputation did it. That is why the sister company matters. Oahu Mold Water Fire launched brand new with no Google Business Profile, no reviews, and no local reputation, and we gave it the foundation only, none of the off-page push. Fourteen days later it was the number eight site the AI was learning from on Oahu and cited by ChatGPT on 38% of non-branded questions. The first lead arrived through Gemini that same window, and the team won the job. Same foundation, no reputation, still on the list. The site itself was doing the heavy lifting.

The mistakes this build avoids

• Leading with off-page work before the site can be read. Reputation compounds on a readable foundation. It cannot substitute for one.
• Trusting a shortcut file to do the work. Publishing llms.txt is fine, but Google has said it ignores the file, so it is not the lever. Crawlable content, schema, and consistency are.
• Writing generic service pages. A machine cannot cite "we're the best in the business." It cites a specific, useful answer to a specific question.
• Letting the Google Business Profile sit incomplete. For local answers it is the primary feed. An outdated profile hands the engines the wrong facts.

Who this method fits

Any home services company with a defined service area and real customers that is simply not legible to machines yet. Restoration proved the hardest version of the problem, an emergency buyer and a national competitor, and the method still worked. Garage door, HVAC, plumbing, roofing, and electrical operators start from the same place and usually face less entrenched competition in AI answers, because most of their competitors have not done this work.

Where it is a longer build

A new business with a thin footprint and no reviews has less for the engines to learn from, so the climb starts slower. Overtaking a deeply established incumbent in a large metro is a matter of months, not the three weeks Kauai took on a smaller island. And we will say plainly what we tell every client: no one can guarantee a fixed position in an AI answer, because the engines change. What is repeatable is the foundation, the measurement, and the climb up the list of sites the AI cites.

Frequently asked questions

Was the 21-day result a one-off? The control test says no. The Oahu company reached the citation list in 14 days on the foundation alone, with no reputation to lean on. Two different markets, same method, same direction.

Does this only work for restoration? No. Restoration was the hard test, an emergency buyer and a national franchise. The same build applies to garage door, HVAC, plumbing, and other home services, which typically face less competition in AI answers today.

How much of this is the website versus reviews and off-page work? The website and structured data are the foundation and move first, which the no-profile Oahu launch demonstrated. Reviews and off-page work compound the result on top of that foundation rather than creating it.

Do we need a Google Business Profile? It is a major help, since it is the primary data feed for local AI answers. Oahu showed you can begin without one, but a complete, accurate profile is one of the strongest signals available.

How do we find out where our business stands today? Advocate 1917 runs an AI Visibility Audit that pulls the real prompts your customers ask and shows which engines name you, which name your competitors, and the gap between. It is the same baseline we ran for Kauai before this build.

The takeaway

A restoration company beat a 2,400-location franchise in AI search in three weeks by being the easiest business for a machine to describe: a readable site, structured data, real answers, and sources that agree. That is a build, not a fluke, and it is repeatable across home services.

See where your business stands in AI search today. Get the AI Visibility Audit at Advocate 1917. For the full playbook behind engagements like this, read The Home Services Marketing Playbook or browse the full playbook library.