Named in AI answers for 11 of 15 emergency questions
A Knoxville restoration company was invisible the moment a pipe burst at 2am. Answer-first emergency content made them the name AI gives.
The situation
Cumberland runs 24-hour emergency response across Knox and Blount counties, which means the phone ringing at 2am is the entire business model.
The owner noticed something: the calls that came in through Google were fine, but he was hearing from more and more homeowners who said they had asked their phone what to do and then called whoever it named. He was not being named.
The problem
Restoration is the purest emergency-intent category there is, and Cumberland had no content that met a person in an emergency.
Their site had a services page, an about page, and a gallery. Nothing that answered what do I do right now. Meanwhile national franchise sites and insurance carrier blogs had published exactly that content, so when a panicked homeowner asked a model what to do about a burst pipe, the model quoted a carrier in Hartford and named a franchise.
They also had no schema, no named author, and no license or certification data anywhere a machine could read it. To a language model deciding who to trust on water damage, an IICRC-certified firm with twenty years in Knox County was indistinguishable from a lead-generation site.
The work
Built the emergency question inventory
Pulled 15 questions directly from after-hours call recordings. What do I do first, do I need to call insurance before or after, will my policy cover this, how fast does mold start. Real panic, real wording.
Wrote answer-first emergency content
Every page opens with a direct 40 to 60 word answer, including the practical first step. No brand introduction, no scroll before the useful part. A model can lift the answer cleanly and a homeowner can act on it in ten seconds.
Published the insurance claim reality
Six pages on what adjusters actually look for, what carriers routinely deny, and where homeowners lose money on restoration claims. Written honestly, including where using their own preferred vendor is the right call.
Made the credentials machine-readable
Full LocalBusiness schema, IICRC certification numbers, named author attribution with credentials, and consistent licensing data across 29 citation sources.
The results
Tracked with monthly prompt testing across four platforms, plus call tracking and CRM job source tagging.
- Cumberland earned citations on 11 of 15 target emergency questions, up from zero at baseline.
- After-hours emergency calls rose 68%.
- Insurance-direct jobs rose 34% month over month average.
- Non-branded organic sessions rose 52% as a byproduct.
Be careful reading these: this window included a January freeze event that produced a burst-pipe surge across all of East Tennessee. Every restoration company in the market had a good quarter. The AI citation count is the cleaner metric, because weather does not change whether a model names you.
Somebody called at three in the morning and said the AI told them to shut off the main and call us. In that order. We had never had that happen before.
In emergency trades, the buying decision happens in under four minutes and increasingly it starts with a question typed into a phone rather than a search for a company. If your site does not answer what do I do right now, you are not in that conversation, and a national franchise or an insurance carrier will be.
Here is the test. Ask ChatGPT what to do about a burst pipe in your city. If a company gets named and it is not you, look at what that company published. It will almost always be a page that answers the question in the first paragraph without selling anything. That is the entire mechanic, and it is available to you this week.
William Burkhart
Founder & CEO, Beacon Primary Care