Blog
We Asked ChatGPT to Pick a Water Softener Company in 8 Cities. Here's What It Looked At.
SEO/AEO
Strategy

We Asked ChatGPT to Pick a Water Softener Company in 8 Cities. Here's What It Looked At.

Read Time:
8 Minutes
Last updated:
November 13, 2026
by 
Caroline Ecklin
Homeowner at her kitchen island researching water softener companies on her phone, next to a glass of water and a water test report.
Table of Contents
Speak With An Expert
Share

When we read through more than 64,000 Google reviews of nearly 900 US water treatment dealers this year, the “how did you find us” answers were what you'd expect: a previous dealer who stopped calling back, a plumber's referral, a realtor, a home show, Yelp. And in a couple of reviews from 2026, ChatGPT.

A couple of reviews isn't a trend. But homeowners don't write “I found you on ChatGPT” in a review unless it happened, and our team wanted to know what they saw when it did. So we asked.

How we ran the test

On November 13, 2026, our team put the same question to ChatGPT, with web search on, for eight metros where hard water is a real sales conversation: Austin, San Antonio, Phoenix, Tampa, Jacksonville, Nashville, Raleigh and Columbus.

“What's the best water softener company in [city], [state]?”

We ran it through a ChatGPT search API set to a US location, so the results weren't shaped by our own history or logins, and we saved each full answer, the pages ChatGPT pulled, and the companies it named.

It was one answer per city, so read what follows as patterns rather than rankings. Ask the same question tomorrow and you'll get a somewhat different list. The patterns are what held across all eight.

1. It answers like a map pack with opinions

Every answer led with four or five named local companies, each shown with a star rating, a review count, an address and a phone number. Then it added its own commentary: “my overall pick,” “worth considering for value-conscious buyers,” “good option for comparing established providers.”

That matters because the homeowner never sees a page of ten blue links. They see five names, a recommendation, and a phone number they can tap. If you're not in those five, you don't get a second look in that conversation.

2. It read the directories, not your website

The source that showed up most was Angi, which ChatGPT drew on in all eight cities. HomeAdvisor, the Better Business Bureau and Thumbtack showed up again and again too.

The second big source surprised us: local “best water softener companies” lists, often published by companies in other trades. In Austin, two of the pages ChatGPT read were ranked lists of water treatment companies written by an HVAC contractor and a plumbing company.

Dealers' own websites were rarely what it pulled from. That doesn't mean your site is irrelevant. It's where a homeowner goes after they've seen your name, and it's what Google ranks when someone searches from the next town over. We looked at that side separately in our research on city pages for water treatment dealers. But in these eight answers, being chosen happened mostly on other people's pages.

3. A small review count didn't keep anyone out. A low rating got called out.

We expected ChatGPT to favor the dealers with the most reviews, and the big review counts were there: several companies with 800 to 1,200 reviews. But so were companies with 3, 9, 12 and 20 reviews, named in the same list.

Dot chart of the review counts of water softener companies ChatGPT recommended in eight US cities, ranging from 3 to about 1,200 reviews, with several companies under 30 reviews named alongside much larger ones.

Ratings were a different story. In three of the eight cities, a local location of a national water treatment brand appeared with a rating below 4 stars, and ChatGPT made the low rating part of the description rather than hiding it. In Tampa it explicitly noted the lower rating next to the name. A brand name didn't protect a location with a weak local reputation.

For an independent dealer, that cuts both ways. You don't need a thousand reviews to be named. But the reviews you do have are being read and summarized for the customer, including the bad ones.

4. It coaches the homeowner before they call you

This was the most useful thing in the test for anyone who sells water treatment. ChatGPT didn't stop at names. Most answers told the homeowner how to buy, and the advice was consistent:

  • Get a real water test, with actual hardness numbers, “not just a sales presentation.”
  • Get the warranty in writing, broken out by tank, valve and installation.
  • Get an itemized quote, and compare at least three companies.
  • Ask about ongoing costs: salt, annual service, filters.
  • Avoid pressure sales.
Five things ChatGPT told homeowners to insist on before hiring a water softener company: a real water test, a written warranty, an itemized quote, ongoing costs, and no pressure sales.

Several answers also gave price ranges. The Austin answer, for example, put a basic whole-house softener at about $1,200–$2,000 installed and a softener with reverse osmosis at $3,500–$6,000 or more.

So a homeowner who arrives through ChatGPT has already been told to test, compare three quotes and watch for pressure. That's not a bad thing for a good dealer. It lines up closely with our review research, where the water test was the moment customers praised most, and the messages they rewarded were written test results, out-the-door pricing and “we'll tell you if you don't need it.” ChatGPT is coaching buyers to expect exactly what the best dealers already do.

Our reviews also show who wins once buyers compare. About 1 in 10 long 5-star reviews say the customer got several quotes, usually three and sometimes as many as 12, and the winner was often the dealer who explained the water best. Words like “explained” and “walked us through” show up in 14.8% of 5-star reviews and 5.1% of 1–2 star reviews.

5. Messy listings showed up as messy answers

In one city, the same company appeared twice in the top five, once with 378 reviews and once with 25, pulled from two different listings with two different addresses. ChatGPT treated them as separate businesses.

You can see how that happens. Old addresses, a second listing someone set up years ago, a different business name on Yelp than on Google. An AI assistant pulling from several directories can't always tell they're the same company, and a split listing splits your reviews.

What we'd do this quarter if we ran a dealership

None of this needs a new “AI strategy.” It's mostly the local basics, with a bit more weight on the places ChatGPT actually read.

  1. Clean up your directory profiles. Claim Angi, HomeAdvisor, BBB, Thumbtack and Yelp. Make the business name, address, phone number and services match everywhere, and merge or close duplicates.
  2. Keep reviews coming, and answer the bad ones. In our review research, dealers who reply to 90% or more of their reviews have 2.5% negative reviews, against 5.7% for dealers who never reply. That's a correlation, not proof, but a low local rating was the one thing that got a company flagged in our test. Owners also reply less to the reviews that matter most: 64% of 5-star reviews get a reply, against 50% of 1-star reviews. Ask after the follow-up retest as well as at install, and reply to every 1–2 star review within a day.
  3. Earn a spot on the local lists. The plumbers, HVAC contractors and realtors in your market are writing “best of” pages. Being the company they actually send customers to is how you end up on them.
  4. Publish what the buyer is being told to ask for. A sample water test report. Your warranty terms broken out by tank, valve and labor. Honest price ranges. What salt and service cost per year. Which control valve you install and whether other dealers can service it. If ChatGPT tells a homeowner to ask for these, be the dealer whose website already answers them.
  5. Ask every customer how they found you, and add “ChatGPT / AI assistant” as a lead source in your CRM. Right now most dealers have no idea whether this is two customers a year or twenty.
  6. Run the test yourself. Once a month, ask ChatGPT the same question for your city and save the answer. It takes two minutes, and it's the only way to know whether you're in the five.

What this test doesn't tell you

Eight answers on one day is a snapshot. It doesn't tell you how many homeowners use ChatGPT to choose a dealer (judging by our review research, still a small share), how ChatGPT weighs one source against another, or whether getting named turns into booked water tests. Results also vary between runs, so a company named once won't necessarily be named the next time.

What it does tell you is where the answer comes from today: directory profiles, review ratings, other people's “best of” lists, and a buyer who has been told to expect a real test and a clear price. Those are things a dealer can work on.

Getting named in AI answers is part of our SEO and AEO work for water treatment dealers. If you'd like us to run this test for your market and see whether you're in the answer, get in touch. We'll show you what ChatGPT says about you before we talk about anything else.

Methodology: on Nov 13, 2026, one ChatGPT answer (web search on, US location, via DataForSEO's ChatGPT search API) per metro for the question above. Review figures come from Matyx's analysis of more than 64,000 Google reviews of nearly 900 US residential water treatment dealers.

"Ranking for water treatment terms is easy. Ranking for the ones that turn into installs is the job."

Caroline Ecklin at Matyx
Caroline Ecklin
CEO

MORE STORIES
FROM THE MATYX

View All
Water treatment technician checking his route sheet at the back of a service van on a suburban street.
Blog

Do City Pages Work for Water Treatment Dealers? We Checked 19 Dealers in 3 Markets

Homeowner on the phone at her kitchen counter, waiting to hear back about her water
Blog

Bad Leads or Missed Leads? How Water Treatment Dealers Can Tell the Difference

Homeowner at her kitchen island researching water softener companies on her phone, next to a glass of water and a water test report.
Blog

We Asked ChatGPT to Pick a Water Softener Company in 8 Cities. Here's What It Looked At.