On November 2, 2026, we scanned the websites of 225 home builders in Houston, Kansas City and Central Arizona. All of them belong to their local builders association and run their sites on WordPress or HubSpot. Most are custom builders, and most are in Houston, so read this as a picture of local builders on the two most common platforms rather than the whole industry.
The short version: the typical builder website is slow, and it isn’t built to sell.
Picture the buyer. They tapped your name from a listing, a friend’s text or an ad, sitting in the car outside a lot. In our phone test, the typical builder homepage took 10.3 seconds to show its main photo or headline. When it arrives, nearly half the sites show no price, offer nothing to do short of calling, and make no specific ask. A family six months from being ready to talk to a builder has one real option: leave. The builder never learns who they were.
None of that is about the houses. It’s about what the website does with the attention the houses earned.
How slow are builder websites on a phone?
Slow. Of 190 homepages, 165 (87%) took longer than 4 seconds to show their main content, the point where Google rates a page poor. 99 took longer than 10 seconds. Only 5 met Google’s 2.5-second mark for good, and the typical homepage weighed 4.1 megabytes. 1 Source 1 Google, “Largest Contentful Paint (LCP),” web.dev: good is 2.5 seconds or less, poor is above 4 seconds. The 10-second line is ours.
Nearly nine in ten builder homepages are slow in our phone test
Mobile lab results for 190 builder homepages: 165 took more than four seconds, Google's poor threshold. Median: 10.3 seconds. We split the poor group at ten seconds to show the longer waits.
- 2.5 seconds or lessgood2.6%5 of 190 · 95% interval 1.1–6%
- Over 2.5 to 4 secondsneeds improvement10.5%20 of 190 · 95% interval 6.9–15.7%
- Poor above 4 secondsOver 4 to 10 secondspoor34.7%66 of 190 · 95% interval 28.3–41.7%
- Over 10 secondspoor52.1%99 of 190 · 95% interval 45–59.1%
Line with end caps: 95% Wilson interval. Rounded shares may not total exactly 100%.
Table view
| Time to main content | Google's band | Homepages | Share of 190 | 95% interval |
|---|---|---|---|---|
| 2.5 seconds or less | good | 5 | 2.6% | 1.1 to 6% |
| Over 2.5 to 4 seconds | needs improvement | 20 | 10.5% | 6.9 to 15.7% |
| Over 4 to 10 seconds | poor | 66 | 34.7% | 28.3 to 41.7% |
| Over 10 seconds | poor | 99 | 52.1% | 45 to 59.1% |
Count ten seconds out loud and imagine a stranger waiting through it for a photograph of a kitchen.
One thing to know before you forward this to your web developer: it’s a lab test on a simulated mid-range phone and a slow connection, three runs per site. A buyer on good Wi-Fi will see something faster, and parts of the page can appear before the main photo does. But it’s the standard Google uses, and the gap between 2.5 seconds and 10 is not a rounding error. 2 Source 2 Largest Contentful Paint under Lighthouse’s standard mobile profile, the one PageSpeed Insights uses. Three runs a site, median kept. A lab test, not real buyers’ phones.
Speed also gets sold as an AI fix. We’ve seen no data, ours or anyone else’s, that ties load time to whether ChatGPT or Gemini recommends a business. Fix it for your buyers.
Where do your plans and available homes send buyers?
The buyer who clicks an available home or a floor plan is your most serious visitor. Where that click lands matters more than almost anything on your homepage.
Available homes. 57 of the sites list homes for sale, and 10 of those 57, about one in six, send the buyer somewhere else to see the home: the Houston MLS site, Zillow, realtor.com, a new-home portal or an agent’s search. One sends all twelve of its homes to the MLS site’s front page, not even to the listings. A page like that can show your buyer other builders’ homes, puts an agent’s form where yours should be, and tells you nothing about who looked. You don’t have to leave the MLS to fix it. Stop Sending Your Best Buyers to the MLS covers keeping the listing and changing the link. 3 Source 3 The scan read each site’s homepage, plan and inventory pages, community pages and three plan pages. Every site it flagged for sending buyers away, or for PDF plans, was checked by hand.
Most of the sites, 135 of 192, list no homes at all. For a custom builder that’s usually the business, not a gap.
Floor plans. 65 of the sites show floor plans, and 11 of those, again about one in six, include plans as PDFs a buyer has to download. Nine of the eleven also have a plan page; for two, the PDF is all there is. One is a scanned image, so its bedroom count and square footage exist only as pixels.
A printable plan is handy, and Google can index a PDF. For the buyer, though, it’s a dead end: no photos beside it, no inquiry form, no way on to the community or the next plan. Nobody searches for “four-bedroom plan with a scullery” hoping for a download. Give each plan its own page with photos and a form, and offer the PDF from there. That page is what buyers search for long before they search for a builder.
Are builders blocking AI assistants?
Almost none are. Only 5 of 192 sites restrict AI in their robots file, and all five do it with a blanket rule that includes Google-Extended, which controls whether Google may use a site to train Gemini and ground its answers. It has no effect on Google Search. None of the 192 block the crawlers ChatGPT, Perplexity or Claude use to find and read pages, and 179 have no AI rule at all. 4 Source 4 Robots files read the way crawlers read them, scoring the crawlers assistants use to find and open pages, not training crawlers. Google-Extended: Google’s crawler documentation.
So if a buyer asks ChatGPT for a custom builder in your county and it names someone else, your robots file is rarely why. What matters is what the assistant finds once it’s in: whether your plans, communities and reviews are on pages it can read and quote. When a Buyer Asks ChatGPT for a Builder covers what decides the answer.
A brochure with a phone number
Now the sales side. Of 186 sites, 138 (74%) have a phone number a buyer can tap. Past that, the options thin out fast:
- 153 (82%) offer nothing for the buyer who isn’t ready to call: no guide, no cost calculator, no newsletter, no plan book.
- 121 (65%) show no price on any page we read: no starting range, no cost guide, no “homes from the 600s”.
- 85 (46%) have no call to action at all, and another 58 (31%) just say “Contact us” or “Learn more”.
Put together, 86 of 186 sites (46%, with an interval of 39% to 53%) miss on all three: no price, no next step and no specific ask. 5 Source 5 95% Wilson intervals throughout. We checked the sales scan by hand on a random 20 sites; 92% of its calls held up, and six measures that held up less well are left out.
What 186 builder websites leave out
What 186 builder websites leave out. The highlighted row is the three gaps together on the same site; the other rows are single findings and overlap.
- No call-tracking script detected90%168 of 186 · 95% interval 85–94%
- No CRM or marketing automation script detected86%160 of 186 · 95% interval 80–90%
- No next step short of contact82%153 of 186 · 95% interval 76–87%
- No dated post found in the last 12 months69%128 of 186 · 95% interval 62–75%
- Inquiry form uses a WordPress plugin66%122 of 186 · 95% interval 59–72%
- No price found on the pages read65%121 of 186 · 95% interval 58–72%
- No qualifying headline above the fold58%108 of 186 · 95% interval 51–65%
- No price, no smaller next step, no specific call to action46%86 of 186 · 95% interval 39–53%
- No current analytics detected41%76 of 186 · 95% interval 34–48%
Line with end caps: 95% Wilson interval. Rows overlap; each is a share of the same 186 sites.
Table view
| What the site leaves out | Sites | Share of 186 | 95% interval |
|---|---|---|---|
| No call-tracking script detected | 168 | 90% | 85 to 94% |
| No CRM or marketing automation script detected | 160 | 86% | 80 to 90% |
| No next step short of contact | 153 | 82% | 76 to 87% |
| No dated post found in the last 12 months | 128 | 69% | 62 to 75% |
| Inquiry form uses a WordPress plugin | 122 | 66% | 59 to 72% |
| No price found on the pages read | 121 | 65% | 58 to 72% |
| No qualifying headline above the fold | 108 | 58% | 51 to 65% |
| No price, no smaller next step, no specific call to action | 86 | 46% | 39 to 53% |
| No current analytics detected | 76 | 41% | 34 to 48% |
That’s the number we’d put in front of a builder’s owner first, because every part of it can be fixed on the website you already have.
What about the buyer who isn’t ready to call?
Most people on a builder’s site are months from a conversation. They’re comparing builders, working out a budget, deciding whether to build at all. A site that offers only a phone number asks them to call before they’re ready. Few will. The rest leave, and you never get their names.
A family six months from being ready to call has one real option on most of these sites: leave.
A next step gives that buyer something to take with them: a cost guide, a plan book, a short checklist of what building with you looks like. In return you get a name and an email address, months before a competitor hears from them.
Price works the same way. Custom builders have good reasons not to post one, and we’re not arguing for a price list. Only 22 of 186 sites have a pricing page of any kind. But a starting range with its assumptions, or a page on what moves the cost, isn’t a quote. It’s a reason for a family to keep reading, and it means the first sales call starts with a buyer who already knows roughly where they stand.
What does your homepage say in the first five seconds?
Most of these homepages open with the builder’s name, a photograph and a city. 108 of 186 (58%) have no real headline at the top, nothing that says what the company builds, for whom, or why this builder. 71 put a slideshow at the top.
We also had each homepage’s opening text read for a five-second test: could a reader tell what you do, for whom and why you? Only 39 of 186 (21%) scored better than halfway. 6 Source 6 The headline rule looks for a main heading of 3 to 14 words above the fold. The five-second test was scored by a model reading the opening text, not by timing real visitors.
Try it on your own site. Show your homepage to someone who doesn’t know your company for five seconds, then ask what you build and where. If they can’t say, neither can a buyer.
The ask matters too. A specific call to action names the next step: “Schedule a design consultation”, “Get pricing on this plan”. Only 43 of 186 sites (23%) lead with one. A specific ask tells the buyer what they’re signing up for, and tells your team what the buyer wanted before anyone picks up the phone.
What happens after someone fills in the form?
On 122 of 186 sites (66%), the inquiry form is a WordPress plugin. On 35 (19%), it’s a CRM’s own form. Neither tells you much by itself. A plugin can forward into a CRM behind the scenes, and a CRM form is only as good as whoever owns what lands in it. We didn’t submit any forms, so we can’t say how fast anyone answers.
You can, though. Four questions for an afternoon:
- Where does a web inquiry actually go: an inbox, a CRM, someone’s phone?
- Who owns it, and who covers when they’re on a site visit?
- Does it arrive with the plan or home the buyer asked about?
- How long until someone replies?
Small frictions add up as well. 70 sites (38%) require a phone number on the form, and 54 (29%) put a CAPTCHA in front of it. Each is a reasonable choice. Each also asks a little more of a buyer who was one click from asking about a home.
Which calls came from where?
The phone is the main way in on most of these sites, and 168 of 186 (90%) had no call-tracking script we could detect. 160 (86%) loaded no CRM or marketing automation script, and 76 (41%) loaded no current analytics we could see. 7 Source 7 These are what the browser loaded. A tag held back by a consent setting, or measurement done on the server or inside the business, would not show.
The ad budget, the open house, the sign on the lot and the website all produce calls. Without call tracking, the website can’t tell you which did. Your team may know from asking, and the next marketing budget gets set on that memory.
Is anyone writing for buyers?
84 of 186 sites (45%) have a blog. On 128 (69%), we found no dated post from the past year. Among the blogs with enough posts to judge, about a third mostly answer buyer questions; the rest are mostly company news.
That’s a missed opportunity. Posts that answer real buyer questions, like what a lot needs before you can build on it, what changes the cost and how long it all takes, are what search engines and AI assistants have the most to work with. They’re also what your sales team explains on every first call.
And the small things buyers notice: 20 sites had at least one broken link among ten we tested, 14 showed a copyright year that has passed, and 4 still showed the placeholder text their theme shipped with.
What to fix first
Most of the sales problems can be fixed on the site you have, without a rebuild:
- A headline that says what you build and where.
- One specific ask, the same on every page.
- A next step for the buyer who is a year out, such as a cost guide or a plan book, and the email address it earns.
- A form that lands with someone who owns it, with the plan or home attached.
- Call tracking, so the next ad budget is set on calls, not clicks.
The structural problems take more. Plans that live only as PDFs and homes that open on the MLS need a page per plan and a page per home on your own site, kept current by a team that can mark a home sold the same day. That’s usually a rebuild question. Will a New Website Cost You Your Google Rankings? covers what a rebuild keeps, and What a Home Builder’s Website Needs lays out the parts, drawn from builder sites we’ve audited and from rebuilding Travars Built Homes’ site.
Speed belongs with the structural work. It’s a job for a developer, and the first question to ask is what’s causing the wait.
To see where your own site stands, the Builder Website Score below asks six questions, one for each part of a builder’s site. Two of them are the ones this report measured: where your plans live, and where a buyer ends up when they click an available home.
How we did this
We ran each site through our free website assessment, a separate phone speed test, a crawl of its plan and listing pages, and a second scan that grades the sales side of the site. Some sites blocked our scanner, so the speed figures cover 190 homepages and the sales figures 186. We checked the plan and listing findings by hand, and checked the sales scan on a random 20 sites. Scripts and forms are what a browser can see, not an audit of anyone’s reporting or follow-up. We submitted no forms and placed no calls.
This is a count of what these sites show from outside. It holds no lead or sales data, so nothing here says a slow site or a missing price cost a particular builder a sale. Every percentage in the charts carries a 95% interval, and the notes below have the details.
The research is free. So is the diagnosis. 60 seconds →