What Actually Changed Between 2023 and 2026
Three years ago, “search” meant a results page with ten blue links, a map pack, and some ads. A homeowner typed a query, scanned titles, and clicked through to compare. That's still happening, but it's no longer the only path. Google's AI Overviews now sit above the organic results for a large share of local service queries, summarizing an answer and naming businesses before anyone scrolls further. ChatGPT and Perplexity have gone from novelty to habit for a meaningful slice of homeowners, especially younger ones who ask “who’s a reliable HVAC company near me that won’t oversell a repair” the way they used to ask a neighbor.
The mechanical difference matters. A traditional search engine ranks pages and lets the user decide. An AI answer engine reads multiple sources, synthesizes them, and hands the user a conclusion, often with two or three business names attached. That's a narrower door. Where ten organic slots used to compete for attention, now three or four AI-cited names get the whole answer, and everyone else becomes invisible in that moment. The competitive stakes went up, not down.
What hasn't changed: the underlying signals AI models draw from are mostly the same ones Google has rewarded for years. Consistent name/address/phone data, real reviews with specific detail, service pages that actually explain the work instead of padding with adjectives, and a site that's fast and easy for a crawler to parse. AI search didn't invent a new rulebook. It raised the bar on execution and added a new consumption layer on top of the old one.
For contractors, the practical shift is this: your site and profiles now have two audiences to satisfy, a human scanning for trust signals and an AI model looking for citable, specific facts. Write for both and you're covered for both. Write only for one and you'll lose ground in the other, usually without noticing until a competitor starts showing up in the AI answer and you don't.
How AI Engines Actually Pick Which Contractor to Name
AI answer engines aren't ranking pages the way a traditional search algorithm does. They're pulling passages and facts from multiple sources, cross-checking them against each other, and then generating a synthesized answer. That process rewards clarity and consistency over polish. A plain-language paragraph that answers “how much does a roof replacement cost in [city]” with a real range gets pulled more often than a vague page that says “contact us for a free estimate” with no numbers at all.
Four things carry the most weight in practice:
- Consistency across the web. Your business name, address, phone number, and service list need to match on your website, your Google Business Profile, and every directory listing. AI models cross-reference these sources, and mismatches erode confidence in all of them.
- Direct answers in plain text. Pages structured around actual questions (“do I need a permit for X,” “how long does Y take,” “what’s the difference between Z and W”) get quoted more than pages built purely around keyword-stuffed service descriptions.
- Review substance, not just star count. A 4.8 average with three-word reviews reads thinner to an AI model than a 4.6 average with reviews mentioning specific jobs, timelines, and outcomes.
- Structured data. Schema markup (Service, FAQPage, LocalBusiness, Review) gives AI crawlers a machine-readable shortcut to the facts on your page, instead of forcing them to infer meaning from unstructured prose.
None of this is exotic. It's the same groundwork that's made SEO campaigns work for two decades, applied with more discipline because the summarization layer punishes vagueness harder than a ranking algorithm ever did. A page that used to limp into position seven on Google can still get zero AI citations, because “ranked” and “quotable” aren’t the same thing anymore.
The Contractor-Specific Risk: Why Trade Businesses Get Hit Harder
Local service businesses face a specific version of this problem that, say, an e-commerce brand doesn't. Homeowners asking AI tools about a roofer, plumber, or HVAC company aren't asking generic questions. They're asking high-stakes, high-trust questions: “is this company licensed and insured,” “do they offer emergency service,” “what’s a fair price for this repair before I get scammed.” AI models are cautious about naming a specific business for anything resembling a safety or financial decision unless the underlying source material is unusually clear and well-documented.
That means a contractor's site carries more evidentiary weight than a typical local business site. A vague “we do it all” homepage with no licensing detail, no service specifics, and no pricing context gives an AI model nothing solid to cite. A site with a clear license number, a specific service area, plain-language answers to the questions homeowners actually ask before hiring, and consistent NAP data across the web gives it something to work with.
There's also a volume problem. Most contractor websites were built years ago by a web shop that hasn't touched them since, or by the owner's nephew, and they've stayed frozen while the discovery layer moved. A site with five pages and a contact form was passable in 2018. In 2026, it's invisible to an AI model that's trying to synthesize an answer from actual content, because there isn't enough content to synthesize from. This is less a marketing problem and more an information-architecture problem: the facts a homeowner (and an AI model) need aren't written down anywhere the machine can read them.
The trades that feel this first are the ones with the highest-stakes decisions attached: roofing (storm damage, insurance claims), HVAC (emergency no-cool calls), and electrical (safety-adjacent work). Homeowners in those categories are already leaning on AI tools to pre-screen before they ever pick up the phone.
What to Fix First: A Practical Priority Order
Most contractors don't need to rebuild everything at once. There's a sequence that gets the biggest-impact fixes done first, based on what AI models weight most heavily and what's fastest to correct.
| Priority | Fix | Why it matters |
|---|---|---|
| 1 | Audit NAP consistency across site, Google Business Profile, and directories | Mismatched facts get discounted by AI cross-referencing before anything else is even read |
| 2 | Rewrite core service pages around real homeowner questions | Plain-language answers are what gets quoted, not adjective-heavy marketing copy |
| 3 | Add Service, FAQPage, and LocalBusiness schema | Structured data is the shortcut AI crawlers use to extract facts reliably |
| 4 | Build out topic depth (multiple pages per service, not one page for everything) | Depth signals expertise and gives AI models more specific passages to cite |
| 5 | Actively request detailed reviews mentioning specific jobs | Review substance outweighs star count alone in trust signals |
Notice what's not on this list: buying ads, running a rebrand, or chasing a viral social post. None of those touch the actual signals an AI model reads when it decides who to name. This is unglamorous, foundational work, and it's exactly why most competitors haven't done it yet. The contractors who fix the boring stuff first are the ones showing up in the answer six months from now.
Typical timeline for competitive terms once the foundation work above is underway: 4-9 months to see meaningful movement, the same range that's applied to organic SEO for years. AI visibility rides on the same underlying authority signals; there's no shortcut that skips the buildout. Trades with less local competition (a niche specialty service, a smaller metro) can see AI citations show up faster; dense metro markets for common trades like roofing or HVAC take longer simply because more competitors are doing the same foundation work at the same time.
AI Search vs. Traditional SEO: What's the Same, What's Different
The honest answer is that AI search optimization (sometimes called GEO, or generative engine optimization) is roughly 80 percent overlap with solid traditional SEO, and 20 percent new discipline layered on top. Understanding which is which keeps a contractor from either ignoring the shift or overreacting to it.
What's the same: technical fundamentals (site speed, mobile usability, crawlability), earning genuine backlinks and citations, keeping a Google Business Profile current, and building topical depth around your actual services and service area. None of that goes away. A slow, thin, poorly structured site was already losing to competitors in 2020, and it's losing worse now.
What's different: the emphasis shifts from “rank for a keyword” to “be the clearest, most specific answer to a question.” That means writing content structured around actual questions rather than keyword variations, being more explicit about facts (license numbers, service areas, pricing ranges, timelines) rather than implying quality through tone, and treating schema markup as a first-class requirement rather than a nice-to-have a developer adds later. It also means reviews matter in a new way: not just as a trust badge for a human, but as source text an AI model reads for specifics.
The businesses that will struggle most are the ones treating this as a separate budget line item, hiring one agency for SEO and a different one for “AI optimization.” The two aren't separable in practice. A page built to rank well on Google in 2026, done properly, already reads well to an AI summarizer. The reverse is also true. Chasing AI visibility as an isolated tactic without the underlying SEO foundation is building on sand.
How to Check Where You Stand Right Now
Before hiring anyone or rewriting anything, it's worth spending twenty minutes finding out where a business actually stands today. This isn't complicated, and most of it can be done without special tools.
Start by asking ChatGPT or another AI tool the exact questions a homeowner would ask: “who’s a good [trade] in [city],” “how much does [common job] cost near [city],” “is [business name] a licensed [trade] contractor.” Read the answers carefully. Does the business get named at all? If it does, is the information accurate, current, and specific, or is it generic and possibly outdated? If a competitor gets named instead, note what their site or profile does that's different: more detailed service pages, more recent reviews, clearer licensing information.
Next, pull up the Google Business Profile and compare the name, address, phone number, and category listed there against the website's footer and the top three directory listings (Yelp, Angi, or trade-specific directories). Even small mismatches (a suite number present in one place and missing in another, an old phone number still live on a directory) create the kind of inconsistency that erodes trust signals across every source at once.
Then look at the actual service pages. Do they answer specific questions in plain language, or do they lean on generic phrases like “quality craftsmanship” and “customer satisfaction” without ever stating a price range, a timeline, or a licensing detail? Pages that could belong to any contractor in any city are the pages AI models skip past.
Finally, check the last ten reviews across Google and any other platform. Are they recent (within the last few months) and specific (naming the job, the crew, or the outcome), or are they old, generic, and thin? This quick pass doesn't require software or a paid tool. It takes a notepad, twenty minutes, and a willingness to read the results honestly instead of assuming the site is fine because it looks fine to a human eye.
What Not to Do: Common Overreactions
The shift toward AI search has produced a predictable wave of bad advice, and contractors evaluating vendors should know what to be skeptical of.
- “AI-optimized” as a standalone product with no SEO foundation. If a vendor is selling AI search visibility disconnected from actual site content, technical health, and review generation, they're selling a layer with nothing underneath it.
- Keyword-stuffing FAQ sections to game AI answers. AI models are trained to detect and discount low-quality, repetitive content. A page with forty near-identical FAQ entries reads as noise, not authority.
- Ignoring Google Business Profile because “AI search is the future.” The map pack and GBP data still feed both traditional search and most AI answer engines. Neglecting it to chase a newer channel undermines both.
- Panic-rebuilding an entire site overnight. A rushed rebuild with unverified claims, thin new pages, and no review foundation can do more damage than the slow, sequenced fix described above. AI models reward established, consistent signals over freshly manufactured ones.
The pattern across all four: mistaking activity for progress. Adding pages, adding buzzwords, adding a vendor's badge to the footer, none of it moves the needle if the underlying facts about the business aren't accurate, specific, and consistent everywhere they appear. The unglamorous audit-and-fix sequence outlined earlier is slower to describe in a sales pitch, but it's what actually changes whether an AI model names a contractor in its answer.
There's also a question of who's doing the work. Some vendors selling “AI optimization” are simply relabeling an old SEO package with new terminology and no changes to the actual deliverables. Ask a prospective vendor to explain, specifically, what changes on the site itself: which pages get rewritten, what schema gets added, how NAP consistency gets audited and corrected across directories. If the answer stays vague or leans on the phrase itself rather than the mechanics behind it, that's a signal to keep looking.