Ask ChatGPT who does what you do. Are you in the answer?
Your customers didn't stop searching. They stopped clicking. A fifteen-minute test to see whether AI names your business.
By Chris Battis
A friend of mine owns a twelve-person home-renovation company. Twenty-two years in business. More five-star reviews than anyone in his county.
Two weeks ago, at his kitchen table, I asked him to type one question into ChatGPT: "best kitchen remodeler near me."
Eight seconds later it answered. Named three companies. Quoted one of their websites directly: pricing range, timeline, the works.
His company wasn't in the answer. Twenty-two years, and as far as the AI his customers are now using was concerned, he didn't exist.
He stared at the screen for a while. Then: "So who decides this?"
- A brochure“Quality craftsmanship”
- Nothing to quote
- Left out
- An answerWhat it costs, how long it takes
- Something to quote
- Named in the answer
The headline stat was wrong. The thing underneath it is worse
In early 2024, Gartner predicted traditional search volume would drop 25% by 2026, eaten by AI chatbots. The marketing world repeated that number for two years.
We can check now. It didn't happen. People are searching more than ever, and Google still owns about 90% of the market.
So you can relax? No. The disruption showed up, just not where the headline pointed.
When Google shows an AI Overview at the top of a search, which it now does on roughly one in five searches, click-throughs to the actual websites collapse by about 61% (Search Engine Land / Graphite, Jan 2026). Publishers have watched a third or more of their Google traffic evaporate. Meanwhile, traffic referred by AI tools is growing at over 500% a year.
Your customers didn't stop searching. They stopped clicking.
The question gets asked. The answer gets assembled from two or three sources. The customer reads it and moves on, or calls the one business the answer named. There's no page two anymore. There's barely a page one. There's the answer, and there's everyone who isn't in it.
The AI your customers use doesn't know your business either
AI gives you generic output when it doesn't know your business. The fix is never a better prompt. It's giving the model what it needs to know first.
Here's the uncomfortable symmetry. The same thing is now true in reverse.
The AI your customers use knows nothing about your business either, except what it can pull from your website. And most small-business websites give it nothing to pull. The real answers (what the work costs, how long it takes, what goes wrong, what to watch out for) live in your head, your phone calls, and your estimates. None of it is written down anywhere a machine can quote.
My friend's website says "quality craftsmanship" and "exceeding expectations since 2004." The competitor that got quoted has a page that says kitchen remodels in their area run $40K–$75K, take 6–9 weeks, and here are the four things that blow the timeline.
One of those is an answer. The other is a brochure. The AI quoted the answer.
Why this is the best news a small shop has gotten in years
I spent years inside HubSpot and Jasper watching how big companies produce content. A single web page goes through strategy decks, brand review, legal review, and three rounds of approvals. Six weeks if nothing goes wrong.
You can publish an answer this afternoon.
And here's what the answer engines actually reward:
- Real questions. You hear your customers' exact questions every day, in their exact words. The big guys guess at them with keyword research.
- Specific expertise. "In twenty-two years of remodels in this county, the permit step is what blows the timeline, and here's why" beats any corporate content team's output, because no corporate content team knows it.
- Local knowledge. A national chain physically cannot write about your town's housing stock, your county's permits, your region's climate. You can't help but know it.
- A person. These engines weigh experience and trust signals. A named owner who has clearly done the work for two decades scores in a way "our team of professionals" never will.
For once the playing field doesn't just include you. It tilts toward you. While enterprise teams schedule meetings about their AI search strategy, the window is open for the people who can move this week.
And the math is forgiving: if your average customer is worth even a couple thousand dollars, one new customer who found you through an AI answer pays for the entire effort.
The five-question test
You can't fix what you haven't measured. This takes fifteen minutes.
Step 1. Write down the five questions customers actually ask you. Not keywords. The real sentences. From your phone calls, your emails, your estimates. ("How much does a kitchen remodel cost around here?" "Can you do it before the holidays?" "Do you handle the permits?")
Step 2. Ask each one in three places: ChatGPT, Google (look for the AI Overview box at the top), and Perplexity.
Step 3. Score it in a table like this:
| Customer question | ChatGPT cites you? | Google AI Overview? | Perplexity? |
|---|---|---|---|
| 1. | ☐ | ☐ | ☐ |
| 2. | ☐ | ☐ | ☐ |
| 3. | ☐ | ☐ | ☐ |
| 4. | ☐ | ☐ | ☐ |
| 5. | ☐ | ☐ | ☐ |
Step 4. Count. Most owners who run this for the first time score zero out of fifteen. That's not a verdict. It's a baseline. And note who is getting cited for each question, because that's who's getting the call you used to get.
Keep the table. Run it again in ninety days. That change, not rankings and not traffic, is the number that matters now.
Start by finding out where you stand.
