Something changed in how construction buyers find vendors, and most contractors haven’t noticed because it doesn’t show up anywhere they look.
It used to work like this: a developer needs a modular manufacturer, or an owner needs a design-build GC, so they search Google, open a few sites, and build a shortlist by hand. Now, a growing share of those buyers open ChatGPT, Claude, or Perplexity instead and type something like “who are the leading modular construction companies serving the Southeast for multifamily projects?” — and get a synthesized shortlist in thirty seconds. Google itself now answers many searches directly with AI Overviews before anyone sees a traditional listing.
If your company isn’t in those answers, you’re not losing a ranking. You’re absent from a conversation you don’t know is happening.
The discipline of fixing this goes by two names — AEO (answer engine optimization) and GEO (generative engine optimization). Strip the jargon and it’s one idea: make your expertise legible to machines that answer questions. Here’s how that works, in contractor language, and what to do about it this quarter.
How buyers actually use AI to build shortlists
Start with the behavior, because it dictates everything else.
Construction buyers were already silent vetters — researching thoroughly before making contact. AI tools supercharge that habit. An owner’s rep can now ask “what should I look for when qualifying a data center GC?” and get a checklist. A developer can ask “volumetric modular versus panelized for a 120-unit project” and get a comparison. And critically, they ask the follow-up: “which companies should I talk to?”
The AI assembles its answer from what it learned in training and what it retrieves from the live web: company websites, industry publications, directories, reviews, forums. Then it names names. The companies that get named share one trait — their capabilities are stated somewhere clearly enough for a machine to repeat with confidence.
That’s the whole game. An AI won’t recommend a contractor whose website says “building excellence since 1987” and little else, because there’s nothing there to repeat. It recommends the firm whose site plainly says what it builds, for whom, where, at what scale, using which delivery methods — because that’s an answer it can quote.
Google’s AI Overviews apply the same pressure from a different direction. When an Overview answers the buyer’s question at the top of the results page, the traditional listings below it get fewer eyes — including yours, even if you rank well. The consolation is that Overviews cite sources, and cited sources still get the click that matters. So the question shifts from “do we rank” to “do we get quoted.” Different question, different work.
Notice what this rewards: substance over polish. A mid-sized contractor with clear, specific, well-structured content can out-answer a bigger competitor with a prettier, vaguer site. That window doesn’t stay open forever.
AEO and GEO in plain terms
AEO means structuring content so answer engines — ChatGPT, Claude, Perplexity, Google’s AI Overviews — can lift your answers and attribute them to you. GEO means the broader work of being visible and credible to generative systems overall: what they know about you, what sources mention you, whether your name comes up when your category does.
Neither replaces SEO. AI systems learn from the same web that Google crawls, so a site invisible to traditional search is invisible to AI too. Think of it as the same foundation with a new floor on top: everything that made construction SEO work — real expertise, clear site structure, authoritative mentions — still applies. What changes is the output format. Instead of ten blue links where being third still gets clicks, an AI gives one synthesized answer with a handful of sources. Fewer winners per question. The companies cited are the ones whose content answered the question best, not the ones with the biggest ad budget — there’s no bidding your way into an answer.
Answer-first content, and why 40–60 words matters
The core AEO tactic is the answer block: a direct, self-contained answer of roughly 40 to 60 words, placed at the top of a page or section, before the background and caveats.
Why that length? It’s the size of a complete spoken answer — enough for a real claim with supporting specifics, short enough to lift whole. Machines pulling text into an AI Overview or chat response work in passages about that size. Give them a clean one and you’re quotable. Make them assemble your answer from fragments scattered across a page and they’ll take a competitor’s cleaner passage instead.
The bigger shift is structural. Most contractor content is written company-first: “Our team brings decades of combined experience…” Answer-first content leads with the buyer’s question and resolves it immediately. What does design-build cost relative to design-bid-build? How long does a modular project take from deposit to installation? What bonding capacity does a project my size require? Answer in the first sentences. Then earn the depth.
Here’s the difference in practice. Brochure version: “At Acme Modular, we pride ourselves on delivering exceptional projects through our proven process.” Answer version: “A typical volumetric modular multifamily project runs 12 to 18 months from contract to occupancy, with modules fabricated while sitework proceeds — the overlap that produces most of the schedule savings.” The second one gets quoted. The first one doesn’t exist, as far as a machine assembling an answer is concerned.
This feels backwards to anyone trained on brochure copy. Do it anyway. It’s also simply better writing — buyers prefer pages that respect their time, and machines are just faster at punishing the ones that don’t.
Where answer blocks belong
Not just blog posts. Service pages should open with a plain answer to “what does this company do and for whom.” Vertical pages should answer “what does this firm know about my project type.” Project pages should state scope, method, and delivery plainly instead of leaving it to captions. Anywhere a buyer has a question, the page holding your answer should lead with it.
Structured data, in plain terms
Structured data — schema markup — is a labeling layer in your site’s code that tells machines exactly what they’re reading. Humans see your project page; schema tells the machine “this is a construction company at this address, serving these regions, offering these services; this page is an FAQ; this is a review with this rating.”
Think of it like the difference between a pile of drawings and a properly indexed submittal package. Same content — one requires interpretation, the other can be processed with confidence. Machines reward confidence with citations.
You don’t need to learn the vocabulary. You need a website where organization schema, service schema, FAQ schema, and location markup are built in rather than bolted on afterward. It’s one of the quiet reasons platform quality matters: structured data baked into every page beats a plugin approximating it.
llms.txt, and the new plumbing
A newer piece of plumbing: llms.txt is a plain-text file at your site’s root written specifically for AI systems — a concise map of who you are, what you do, and where your most important content lives. Where robots.txt tells crawlers what they may access, llms.txt tells AI crawlers what actually matters. Adoption is early and standards are still settling, but the cost is trivial and it signals machine-readability across the board.
We build this layer — llms.txt, schema, answer-block architecture, AI-crawler-friendly rendering — as standard in our AI systems work, because retrofitting it costs more than including it from day one.
What you don’t control: the rest of the web
GEO has a second half that lives off your website, and it matters more than most guides admit.
AI systems triangulate. Before naming your company in a shortlist, they weigh what the rest of the web says: industry publications, association directories, project announcements, podcast appearances, reviews. A company described consistently across several independent sources gets named with confidence. A company that exists only on its own website is a claim without a witness.
For contractors, this translates into unglamorous but concrete work. Keep your Google Business Profile complete and current. Get your project wins written up in trade press, even briefly. Make sure directory listings say the same thing your website says — same specialties, same regions, same company description, so machines see agreement instead of noise. If you or your leadership speak on industry podcasts or panels, get that on the record online. None of this is new advice. What’s new is that machines now read all of it at once and score the consistency.
Why FAQ pages matter again
For years FAQ pages were an afterthought. AI search made them prime real estate, for a simple reason: answer engines answer questions, and an FAQ page is questions and answers in exactly the format machines lift.
Every real question a buyer asks — about financing, timelines, delivery methods, service areas, what happens after they submit an inquiry — is a chance to be the quoted source. Add FAQ schema and each entry becomes individually machine-readable.
Two rules. First, real questions, phrased the way buyers actually ask them — pull them from sales calls, emails, and preconstruction meetings, not from a brainstorm about what sounds good. Second, direct answers: first sentence answers the question, the rest adds context. An FAQ that dodges into marketing copy is worse than none, because it teaches machines your site doesn’t answer things.
What to do this quarter
This is a quarter’s worth of focused work, not a rebuild of your business.
- Run the test. Ask ChatGPT, Claude, and Perplexity the questions your buyers would ask — “best [your specialty] contractors in [your region]”, “how to choose a [your category] company”. See who gets named. That’s your real competitive set in AI search.
- Fix your foundational answer. One clear passage stating what you build, for whom, where, at what scale, with which methods — on your homepage and about page, quotable in under 60 words.
- Rewrite your top pages answer-first. Services and vertical pages: lead with the answer, follow with proof.
- Build the FAQ layer. Twenty real buyer questions, direct answers, FAQ schema.
- Add the technical layer. Organization and service schema, llms.txt, and confirm AI crawlers aren’t blocked.
- Keep publishing. One genuinely useful answer to a real buyer question each month compounds; machines keep reading long after launch week.
We track how AI systems describe construction companies and what they cite on our AI resources page, and the pattern is consistent: specific beats general, structured beats scattered, and early beats late.
Your buyers are already asking the machines. The only question is whose answers come back.