How to Get Your New Home Community Into ChatGPT and Google AI Answers

A buyer moving to Charleston or Huntsville used to type "new homes near me" into Google and get ten blue links. More and more of them now ask ChatGPT, Gemini or Google's AI Mode, "Which home builders should I look at?", and get back a short list of names.

If your company isn't on that list, you never get the click. So in September my team and I tested it. We didn't run a survey. We asked the assistants directly and recorded what they said.

What we tested

As part of our study The Lead Response Gap 2026, we asked three AI assistants, each with live web search turned on, the same three questions a buyer might ask, for each of 20 Sun Belt metros:

1. "What are the best new home builders in [metro]? Name specific builder companies."
2. "I want to buy a new construction home in the [metro] area. Which home builders should I look at, and why?"
3. "Which local or regional home builders (not the big national ones) build new homes around [metro]?"

That's 180 answers, all collected on September 27, 2026, from ChatGPT (gpt-5.5), Claude (claude-opus-5) and Gemini (gemini-3.8-flash with Google Search grounding). We then checked which of the 160 regional builders in those metros (builders marketing two or more selling communities) were named.

Half of regional builders never came up

• 52% were named at least once by at least one assistant.
• ChatGPT named 46% of them, Claude 37% and Gemini 37%.
• Only 28% were named by all three.

So for roughly half of the regional builders in those markets, a buyer who asks an AI for options never hears their name. And the assistants don't agree with each other: being on ChatGPT's list doesn't put you on Gemini's.

What the named builders had in common

We had already measured each builder's website from the outside: whether it shows a lot map, publishes prices, gives buyers a way to get an answer after hours, and carries visible ad measurement (a Google Ads tag or call tracking). We put those signals into a statistical model alongside builder size and state.

Logistic model with state controls, n = 150. Source: The Lead Response Gap 2026, Exhibit 6.

Two things stood out. Builders that showed a lot map, and builders that measured their ads, were named about three times as often, even after accounting for size. And publishing a price, which I expected to matter, showed no clear effect.

Now the honest part. This is an association, not proof. Better-run builders tend to do everything well, and they get talked about more. Adding a lot map tomorrow won't make ChatGPT recommend you on Friday. The answers also change from one run to the next, so this is a snapshot of one day. We ran one more check on the 42 builders where we could also see Google ad activity, and it pointed the same way (measurement 5.3×, with a wide interval of 1.1–25).

My reading as an operator: a lot map and clean measurement are signs that a builder runs its marketing as a system. A website that shows buyers exactly what is for sale, lot by lot, is also a website a machine can read.

Where the AI got its answers

This part isn't in the published report. We went back to the same 180 answers to see which sources the assistants used.

• ChatGPT cited about 12 different websites per answer, and roughly seven in ten were builders' own websites. Your site is the main raw material.
• The third-party sources that kept coming back were NewHomeSource (cited in 18 of ChatGPT's 60 answers), Builder magazine's Local Leaders lists (13 of 60), Houzz, local home builders association and Parade of Homes sites, and regional award lists.
• In Claude's searches, NewHomeSource came up in 59 of 60 answers and Houzz in 56. Yelp, Wikipedia, Zillow, LinkedIn and the BBB appeared regularly too.
• Gemini doesn't show its sources in a form we could read (its links are Google redirects), so we can't say what it relied on.

A caveat on method: we sorted thousands of links into "builder website" and "third party" by domain, which is rough. The pattern is clear even so.

What to do about it

None of these is a trick. Each one follows from what the assistants actually read.

1. Make sure the assistants can read your site

Each AI company sends its own crawler, and your robots.txt file decides whether it gets in.

• ChatGPT: OpenAI says sites that block its search crawler, OAI-SearchBot, won't be shown in ChatGPT search answers (OpenAI).
• Claude: Anthropic says blocking Claude-SearchBot or Claude-User may reduce your visibility in Claude's search results (Anthropic).
• Google AI Overviews and AI Mode: Google says your page only needs to be indexed and eligible to show a snippet in regular Search. There are no extra technical requirements and no special structured data (Google Search Central).

Also check the firewall. In our study, 58 builder sites answered our automated visit with a bot challenge. We don't know whether those sites block AI crawlers too, but ask whoever runs your website to check.

2. Put what you sell on the page, in plain text

If seven in ten of ChatGPT's sources are builder websites, the words on your community pages are what the AI quotes. Every community page should say, in text and not only inside an image or a PDF:

• the city or area, and the nearest landmark buyers know;
• the price range ("from the $380s");
• home sizes and bedroom counts;
• how many homes are available or move-in ready, and when;
• a lot map, interactive or static, that shows what's available.

That last item is the same signal from the table above. A lot map tells a buyer, and a machine, that you have real homes for sale right now.

3. Be where the assistants look

You can't control what an AI says about you, but you can control whether you're on the sites it reads:

• a complete, current NewHomeSource listing for every community;
• home builders association membership and Parade of Homes entries where you build;
• a Houzz profile with real reviews, plus a Google Business Profile and a BBB listing kept up to date;
• local press and awards. Builder's Local Leaders list ranks the top 10 builders by closings in each of the 50 largest new-home markets, using a survey builders fill out each spring (Builder). You can't buy a spot, but if your closings qualify, submit your numbers.

4. Measure your ads

The strongest signal in our model was visible ad measurement. I don't think an AI checks your tracking tags. My guess is that builders who measure what their ads produce tend to be the ones that grow, get covered and get reviewed. Either way, if you spend on ads and can't say which ones produced tours and contracts, fix that first, for reasons that have nothing to do with AI.

5. Study the builders the AI keeps naming in your market

Local search for new homes works differently from most local businesses. A buyer isn't comparing every plumber within five miles. They're choosing from a short list of builders in one metro, and in our test the assistants built that list mostly from builder websites and new-home sites like NewHomeSource, not only from map listings and reviews. You're competing against the same ten or twenty names, and you can see exactly what those names have.

So run the three questions above in ChatGPT, Gemini and Claude for your metro, with web search on. Use the third question ("local or regional, not the big national ones") because that's the list you're actually competing for. Note which builder shows up in every answer. Then compare their website with yours on five things:

1. A lot map. Does every community page show available homesites, interactive or static? This was one of the two strongest signals in our model (3.3×).
2. Ad measurement. Do they advertise, and do they measure it? You can see their ads in Google's Ads Transparency Center, and a free tag checker such as Google Tag Assistant shows whether the page carries a Google Ads tag or call tracking. This was the other strong signal (3.5×).
3. Facts in plain text. Can you select the city, price range, home sizes and available homes on their community page with your mouse? If you can, so can a crawler. On your own site, check whether the same facts are stuck inside an image, a PDF or a map widget.
4. The sources the AI cites for them. Click the links in the answer. If the assistant names them from NewHomeSource, a Parade of Homes page, a Houzz profile or a local "best builders" article, that's where you need to be too.
5. An after-hours path. Can a buyer chat, text or book a tour at 9 p.m.? In our model this pointed the same way (2.1×) but wasn't conclusive, so treat it as a tiebreaker, not a rule. It matters for your sales anyway.

Whatever they have that you don't is your to-do list. Repeat the check once a month and keep a simple log. AI answers change from run to run, so look for the trend, not one answer.

What not to buy

Be careful with anyone who promises to "get you into ChatGPT." Google says there is no special optimization for its AI features, and no one outside these companies controls what their assistants recommend. What works is slower and less exciting: a site machines can read, clear information about what you sell, and a real presence on the sites buyers and assistants already trust.

If you hire someone to help, whether an agency, your web developer or an "AI SEO" vendor, ask them specifically:

1. Bot blocks. Will you check whether our server, hosting or firewall (security plugins, bot protection) blocks the AI crawlers, and show us the result?
2. Indexing and analytics. How will you confirm that every community page is indexed in Google Search Console and Bing Webmaster Tools? And how will we see, in our analytics, the visits that come from AI assistants?
3. Structured data. What will you fix in our structured data, and how will you keep it matching what's visible on the page?
4. Listings in plain text. How will you put our homes, prices and available lots on the page in text a crawler can read, while the website keeps looking the way we want?
5. Profiles and presence. What's the plan for our NewHomeSource listings, Google Business Profile, Houzz, home builders association pages and reviews? Who updates them, and how often?

If the answer to any of these is vague, you're buying a promise, not a plan.

FAQ

Does schema markup get my community into AI Overviews?

Google says no special structured data is required for its AI features. Structured data is still good practice, as long as it matches what's visible on the page.

Do AI assistants favor the big national builders?

They name them often, but in our test they also named about half of the regional builders, and size explained only part of who got named.

How often do AI answers change?

Often. The same question can return a different list on a different day, which is why one check means little and a monthly log means more.

Should I block AI crawlers to protect my content?

That's your call, but blocking OpenAI's or Anthropic's search crawlers means you can't appear in their search answers. Training crawlers, such as Anthropic's ClaudeBot, are separate and can be blocked on their own.

Methodology: The Lead Response Gap 2026, Enroke Research. 160 regional builders with two or more selling communities in 20 Sun Belt metros; 180 answers from three AI assistants with web search, collected September 27, 2026. Odds ratios from a logistic model with state controls (n = 150). Associations, not causes. The report does not name individual builders, and neither does this article. Read the full report.