How ChatGPT Picks Which HVAC Company to Recommend

A homeowner's AC dies on a Saturday in July. Five years ago she typed "AC repair near me" into Google. Today, there is a growing chance she asks ChatGPT instead: "My AC stopped working and it's 90 degrees, who should I call in Columbus?"

ChatGPT answers with two or three company names. Not ten blue links. Not a map with dozens of pins. Two or three names, presented with the confidence of a knowledgeable friend.

If your company is one of those names, you just got the highest-trust referral in marketing. If it is not, you were invisible in a channel you probably are not even measuring. So it is worth understanding exactly how the names get picked, because it is not random, and it is not the same as Google's rankings.

Where AI Recommendations Actually Come From

When ChatGPT gets a local service question, it does not consult some private database of good contractors. It retrieves live information from sources it trusts, then synthesizes an answer. Recent studies of AI citations for local business queries show a consistent pattern:

Two findings in that data should stop HVAC owners cold.

First, Yelp receives more AI citations than every competing local platform combined. Plenty of contractors wrote Yelp off years ago. The AI engines did not.

Second, ChatGPT retrieves through Bing's index, not Google's. A company that ranks well on Google but is thin in Bing's view of the web, weak Yelp presence, no editorial mentions, inconsistent citations, can be strong in the Map Pack and absent from AI answers. Different pipes, different winners. That is one of the patterns showing up in research on what AI cites for local queries.

The Overlap With Google, and the Gap

Here is the useful nuance: most of what wins the Google Map Pack also feeds AI visibility. Reviews, real service pages, consistent citations, authority links, an active profile: these are inputs to both systems. We covered the Google side in depth in our AI Overview visibility guide, because Google's own AI answers pull heavily from the local pack and its trusted sources.

But the systems are not identical, and the gap is where opportunity lives:

What Actually Gets an HVAC Company Recommended

Pulling the retrieval data together, the playbook looks like this, in priority order:

1. A website that answers questions the way AI wants to quote them

Clear service pages, real location pages, plain answers to real questions (cost, timing, emergency availability), with schema markup so machines parse it cleanly. This is 35 percent of the citation pie and fully under your control. It is also, not coincidentally, exactly what our content and technical SEO work builds.

2. A live, well-reviewed Yelp profile

Claimed, complete, photos, and a steady trickle of reviews. You do not need to love Yelp. You need to exist well on it, because the AI engines cite it more than everything else combined.

3. Review volume and recency everywhere

AI assistants echo consensus. A company with recent, positive, detailed reviews across Google, Yelp, and BBB reads as the safe recommendation. Recency matters to AI for the same reason it matters to Map Pack rankings: stale signals read as risk.

4. Editorial and community mentions

Local news pieces, chamber and association pages, supplier spotlights, and yes, Reddit threads where real people say "we used these guys, they were great." You cannot fake the Reddit layer, but you can earn it, and one honest thread can echo through AI answers for years. This is link building wearing its 2026 clothes: the mention now matters as much as the link.

5. Consistent citations

The boring layer still counts: BBB, Foursquare, the trade directories, all agreeing on your name, address, and phone. Disagreement reads as unreliability to machines deciding whether to put their credibility behind recommending you.

Why This Matters Now and Not "Someday"

Google's AI Overviews already sit above organic results for a growing share of HVAC queries. AI Mode is rolling out across Google Search. And the brand-check behavior, a homeowner asking "is [company] any good?" after getting a quote, is already happening to your company today, with an answer being generated from whatever the machines can find about you. Consumer adoption of AI for local business discovery is climbing fast.

Almost no HVAC agency is measuring any of this. Most cannot answer question 11 on our agency interview checklist. We build GEO into campaigns from the start because the same work that wins the Map Pack, done properly and structured properly, is what gets you named by the machines. Two channels, one body of work.

Want to know what AI currently says about your company? That check is part of our free Map Pack audit: your grid, your citations, and what ChatGPT and Google's AI actually say when someone asks about HVAC in your city.

Frequently Asked Questions