Open ChatGPT. Type: "Who's the best roofer in Dallas right now?"
If your company isn't in the answer, you're not in the conversation. And the conversation has already started — without you.
This is the part most agencies in the trades aren't talking about yet. While everyone is still arguing about Google rankings and Facebook ads, a quiet shift is happening in how homeowners find contractors. The shift is from search engines to answer engines — AI tools that don't just list ten links and let you pick. They name names. They recommend specific companies. They cite specific articles. And the companies they recommend are the ones who've already done the work to be cited.
That work is called GEO — Generative Engine Optimization. This is the complete guide. Written for the owner who's smart but doesn't have time to read 30 white papers to figure out what to actually do on Monday morning.
1. The Shift
For 25 years, finding a contractor meant the same thing: open Google, type your need plus your city, scan ten blue links, click two or three, call the one that looked most trustworthy.
That model is still here. It's just no longer the only model.
What's changed in the last 18 months
-
Google AI Overview now appears at the top of millions of search results — including high-intent commercial queries like "best HVAC company in Plano" and "roof replacement cost Texas." The AI Overview pulls from across the web and writes a summarized answer. Sometimes it names specific companies. Sometimes it links to industry blog posts. The blue links you used to compete for? They're now below the AI box, and a meaningful percentage of users never scroll past it.
-
ChatGPT has search built in. Ask it "best plumber in Frisco TX" and it'll return a real answer — typically with a few specific company names, pulled from across BBB profiles, Reddit threads, news articles, and the companies' own websites. SearchGPT capabilities, originally launched as a separate product, are now native to ChatGPT itself.
-
Perplexity is the cleanest example of the new model. Every answer cites its sources transparently. If you ask "compare three roofing companies in Dallas," it'll do exactly that — and the three it picks are the three whose web footprint matches the query best.
-
Google Gemini, Microsoft Copilot, Claude, and Apple Intelligence are all doing variations on the same theme. The answer engine is becoming the default surface.
What this means for your phone
A homeowner whose AC dies on a Tuesday night used to type "AC repair near me" into Google. They still do that. But they also — increasingly — ask Siri, ask ChatGPT, ask Google's AI assistant. And the answer they get isn't a list. It's a recommendation. One company, sometimes three, named by name.
If the AI tools recommend you, your phone rings. If they don't, you weren't even in the consideration set. The homeowner didn't choose someone else. They never knew you existed.
One example to make this concrete
In late 2025, we ran a test query across ChatGPT, Perplexity, and Google AI Overview: "Who's the best roofing company in Plano TX for storm damage?"
Across the three tools, eleven different companies were mentioned. Three appeared on all three platforms. Six appeared on two of three. Two appeared on only one. Dozens of legitimate Plano roofers, including some doing $5M+ in annual revenue with strong Google rankings, appeared on zero.
The companies cited weren't necessarily the biggest. They were the ones whose web footprint made them easy for AI to find, summarize, and trust. That's the entire game.
2. What Is GEO
GEO is the work of making your business legible to AI search tools so they recommend you.
That's it. That's the whole concept.
If SEO is "get on page 1 of Google so a person clicks your link," GEO is "be the company an AI mentions by name when a person asks a question." Different surface. Different signals. Different playbook.
Why "Generative" engine optimization
The term comes from how these new tools work. They don't index and rank pages the way Google's original algorithm did. They generate answers by reading across hundreds of sources and synthesizing a response. The signals that get you generated into the answer are different from the signals that get you ranked on the SERP.
You'll see other names for the same thing — AEO (Answer Engine Optimization), LLMO (Large Language Model Optimization), AI SEO. They mean the same thing in practice. We use GEO because it's the term that's stuck in 2026 and it accurately describes what the optimization is for.
How GEO and SEO relate
Good news: most of GEO is built on top of good SEO. The companies winning at GEO are almost always also winning at traditional SEO. The same content, schema, citation graph, and trust signals that rank you in Google's blue links are the ingredients AI tools pull from.
But there's a layer on top — a set of GEO-specific moves that traditional SEO doesn't cover. That's what most of this guide is about.
If you're starting from a weak SEO foundation, fix that first. GEO won't save a broken local search game. Read our 2026 Roofing Marketing Playbook or HVAC Marketing System for the SEO foundation. Come back here when those are running.
3. Why It Matters for Trade Businesses Specifically
GEO matters more for trades than it does for almost any other industry. Three reasons, all of them about how homeowners actually behave.
1. The emergency moment
A homeowner whose pipe just burst at 11pm isn't in a careful research mood. They're in a panic mood. They want one recommendation, fast, from somewhere they trust. AI search tools are positioning themselves as exactly that — the trusted advisor for high-stress, low-time decisions.
Watch what happens over the next 24 months. Smart speakers, voice assistants, and conversational AI on phones are going to capture more of that emergency moment. "Hey Siri, my AC isn't working — who should I call?" The plumber, roofer, HVAC company, or electrician that wins the AI recommendation wins the emergency.
2. The research moment
A homeowner thinking about a $20K roof replacement spends weeks on research. They Google. They read. They ask friends. And increasingly they ask AI: "How much does a new roof actually cost in Texas?" "Should I file an insurance claim for hail damage?" "What's the difference between architectural shingles and three-tab?"
When AI answers those research questions, it cites sources. The companies whose blog posts, FAQ pages, and capability content get cited get something more valuable than a backlink — they get a recommendation embedded in the answer the homeowner is using to make their decision. And then, when that homeowner is ready to call, they remember the name they saw cited as the source.
3. The reputation moment
"Best HVAC company in Fort Worth." "Top roofers near Plano." "Reliable plumber in McKinney." These are pure shortlist queries — the homeowner has a need, now they want to know which 2–3 names to consider.
These queries used to mostly route through Google and Yelp. Now they increasingly route through AI. Whoever the AI puts on the shortlist gets called. Whoever it doesn't, doesn't. There is no participation trophy.
For trade businesses, this means GEO is not optional, but the window to get ahead is wide open. Most of your competitors haven't started. Almost none of the agencies serving the trades have built it into their service mix. First mover here compounds for years.
4. How AI Search Picks Who to Recommend
Nobody outside the labs at OpenAI, Anthropic, Google, and Perplexity knows the exact algorithms. Anyone who claims otherwise is selling something. What we do know — from observation, from public documentation, and from running thousands of test queries — is that AI search tools weight specific kinds of signals heavily.
These are the seven signals that matter most:
Signal 1: Authoritative, in-depth first-party content
AI models build their understanding of your business primarily from what's on your own website. A roofing company whose About page is two sentences ("Family-owned since 2014, serving DFW") doesn't give the model enough to work with. A roofing company whose About page is 800 words — covering founding, leadership, certifications, service area, equipment lines, capabilities, awards, and differentiators — gives the model a lot to summarize.
Same for service pages. Same for FAQ pages. Depth and clarity beat keyword stuffing every time in GEO.
Signal 2: Structured data (schema markup)
Schema markup is the technical layer that tells AI what kind of entity you are and how to interpret your content. LocalBusiness schema tells it you're a local business with a name, address, phone, service area, opening hours, and rating. Service schema tells it what services you offer. FAQPage schema tells it which questions you answer and where.
AI tools heavily prefer to recommend businesses whose entities they can parse confidently. Schema is how you give them that confidence. (We get into the specific schema types in the playbook section.)
Signal 3: Brand mentions across the web (citation graph)
This is the GEO signal most agencies don't talk about because it's the hardest to manufacture. Your "citation graph" is the network of places across the web that mention your business by name — BBB profiles, Yelp, Angi, Houzz, manufacturer locators, local news articles, Reddit threads, trade publications, chamber of commerce listings, industry association directories.
AI models use this graph to validate that a business exists, is reputable, and is what it claims to be. A business with 200 high-quality citations across diverse sources will be recommended more confidently than one with 20 thin profiles. Citation graph also affects which businesses an AI picks first when forced to choose 3 from 20 candidates.
Signal 4: Consistent NAP data
NAP = Name, Address, Phone. If your name is "Lone Star Roofing" on your website, "Lone Star Roofing LLC" on BBB, and "Lone Star Roof Co." on Angi, you've fragmented your entity. AI models can usually figure it out — but they're less confident, and confidence affects recommendation likelihood.
Same with phone numbers and addresses. Use one canonical version of each across every property you control or claim.
Signal 5: Review volume and quality
Reviews aren't just for showing potential customers you're trustworthy. AI tools also factor them into recommendations. A roofing company with 412 Google reviews at 4.7 stars gets surfaced more often than one with 38 reviews at 4.9 — because the volume validates the rating. AI tools, like humans, get suspicious of small sample sizes.
Multi-platform reviews (Google + BBB + Facebook + Nextdoor) also strengthen the signal. Concentrated reviews on a single platform look less robust than distributed reviews across several.
Signal 6: Topical authority
If your site covers one topic — "roofing" — deeply, across many pages, AI starts to treat you as an authority on that topic. The relationship between topical depth and AI citation likelihood is meaningful. A roofing company with 25 pages on different aspects of roofing (replacement, repair, storm damage, materials, financing, insurance claims, by city, by neighborhood) is interpreted as an authority. One with 5 thin pages isn't.
This is where the "content hub" strategy pays off twice — once for traditional SEO ranking, again for GEO citation.
Signal 7: Recency
AI models care about how current your content is. A blog post from 2022 with current information will be cited less often than a blog post from 2026 covering the same topic. Stale About pages and service descriptions hurt. Frequently updated content helps.
This isn't an excuse to rewrite everything constantly — but content with a clear "updated [date]" notice and genuine current information beats forgotten content from three years ago.
5. The GEO Playbook: 10 Specific Actions
This is the operational section. Ten things to actually do, in roughly the order we'd do them for a new client engagement.
Action 1: Implement comprehensive schema markup
The five schema types every trade business website needs:
- LocalBusiness on the homepage — name, address, phone, service area, opening hours (including emergency 24/7 if you offer it), price range, aggregate review rating
- Service on each service page — service name, description, area served, price-from where you can be specific
- FAQPage on FAQ-heavy pages — each Q/A pair structured, with definitive answers in the first paragraph
- Review for individual customer reviews displayed on the site
- Product for equipment you install (shingle brands, HVAC equipment, water heater brands) — name, manufacturer, photos
Implementation is via JSON-LD in the page's head. Most modern CMSs (and any developer) can deploy this in a few hours. Validate at Google's Rich Results Test before shipping.
Action 2: Expand your About / Capabilities content
Most trade business About pages are two paragraphs. That's not enough for AI to build a rich entity description from. Push it to 800–1,500 words. Cover:
- Founding year and origin story
- Leadership names and bios
- Certifications and licenses (with numbers where applicable)
- Service area (specific cities, not "DFW")
- Equipment brands you install
- Differentiators (24/7 service, dealer tier, financing options)
- Awards and recognition
- Community involvement
Then create capability pages for specific specialties. "Heat pump installation" deserves its own page if you do heat pumps. "Insurance claim assistance" deserves its own page if you do storm work. Each capability page deepens AI's understanding of what you do.
Action 3: Build a 20–30 question FAQ hub
This is the single highest-leverage GEO move you can make in 2026. The questions homeowners ask AI tools are remarkably consistent — and they map directly to the content you should be publishing.
For HVAC, those questions look like: "How much does a new AC cost in 2026?" "Is a heat pump worth it in Texas?" "What's the R-454B refrigerant transition?" For roofing: "How much does a new roof cost in Texas?" "Should I file an insurance claim for hail damage?" For plumbing: "How much does slab leak repair cost?" "Tankless vs. tank water heater?"
Format matters. Each answer:
- Lead with a clear, definitive answer in the first paragraph
- Follow with supporting detail, context, or caveats
- Mark up with FAQPage schema
- Include specific numbers, ranges, or examples where defensible
- Keep the writing tight — AI tools quote concise answers more often than rambling ones
Action 4: Build the citation graph
The unglamorous work that nobody else is doing. The list to systematically claim and complete:
- BBB (full profile, photos, business hours, services list, awards)
- Angi, HomeAdvisor, Thumbtack (all completed)
- Yelp (claimed even if you hate it — Apple Maps and some AI tools pull from it)
- Houzz, Nextdoor (verified business profile)
- Industry-specific directories: GAF Master Elite, Owens Corning Platinum, Trane Comfort Specialist, Lennox Premier Dealer, NATE-certified contractor list, ACCA, PHCC for plumbing
- Local chamber of commerce
- State and regional trade associations
- Google Business Profile (completed to 95%+)
Every citation listing should use the exact same NAP. Inconsistent NAP is one of the most common quiet GEO bugs we find.
Action 5: Format content for AI extraction
AI tools have strong preferences in how content is structured. The patterns that get cited more often:
- Clear headings (H2/H3 hierarchy) that name what each section is about
- Lists and tables rather than dense paragraphs for comparable information
- Definitive opening sentences for each section ("The average roof replacement in Dallas costs $9,500–$18,000.")
- Specific numbers, ranges, and dates rather than vague qualitative claims
- Direct question/answer formatting in FAQ content
- Comparison tables for "X vs. Y" content
Rewrite your top 20 pages with this in mind. Read each page out loud. If the first sentence of each section doesn't deliver a clear, summarizable statement, rewrite the section.
Action 6: Get cited by local news and trade publications
Local TV news (CBS, NBC, ABC affiliates) and local newspaper websites are weighted heavily by AI tools as authoritative sources. So are trade publications (Roofing Contractor magazine, Contracting Business, ACHR News, Plumbing & Mechanical).
In Texas, the cycle is predictable:
- March–April: local news covers storm preparation
- July–August: local news covers heat wave / AC strain stories
- November–December: local news covers winter freeze / pipe protection
- After major weather events: local news scrambles for expert quotes
Pitch yourself in advance of those windows. Call assignment desks. Offer expert availability. Even one quote in a local CBS DFW segment becomes citation fuel for years.
Action 7: Build the review velocity, not just the count
For GEO citation, AI tools care about both the total volume of reviews and the recency. A business with 200 reviews where the most recent is two years old is interpreted as "was good, may not be anymore." A business with 200 reviews where 15 came in the last month is interpreted as "actively operating, currently reviewed."
Build automated review request systems (Podium, Birdeye, NiceJob, or CRM-native). Set a velocity target — minimum 8 new Google reviews per month, 15+ for healthy growth, 30+ for top-quartile. Respond to every review within 24 hours, personally.
Action 8: Implement an llms.txt file (cutting-edge)
llms.txt is an emerging standard for telling AI crawlers which content on your site is most useful for them to index. Think of it as a sitemap specifically for LLMs. It lives at yourcompany.com/llms.txt and lists your most important pages with brief descriptions.
Not every AI tool respects it yet. Some do. The cost to implement is trivial (a single text file). The upside is being early on a signal that is likely to grow in weight. We add it to every client engagement.
Action 9: Monitor your AI visibility
You cannot improve what you do not measure. Build a defined set of 8–12 test queries that matter to your business — "best [trade] in [city]," "[trade] near me [city]," "[specific service] cost [state]," etc. — and run them monthly across:
- ChatGPT
- Perplexity
- Google AI Overview
- Microsoft Copilot
- Google Gemini
Track who appears, who doesn't, and how that changes over time. This is your GEO scorecard. Without it, you're flying blind.
Action 10: Build content that answers questions completely
The single content pattern that wins in AI search: answer the question fully so the AI doesn't have to look anywhere else.
Most contractor content tries to tease — "wondering about roof replacement cost? Call us for a free estimate!" AI tools don't cite teases. They cite complete answers.
A 1,500-word article that walks through real cost ranges, the factors that drive the range, what's included, what's excluded, and how to get an accurate estimate — that gets cited. A 400-word article that gestures at all of that and then asks for a phone call doesn't.
The fear is that complete content will reduce the need to call you. The reality is the opposite — complete content positions you as the expert, and homeowners call experts.
6. MCANIX's Approach
Brief, because this guide isn't a sales pitch.
We build GEO into every client engagement from day one. We've been doing it since we founded the agency in 2026 — long before most of our competitors started talking about it. We do it because we think it's where the puck is going, and because the first-mover compounding for trade businesses is too good to leave on the table.
Practically, that means every new client engagement starts with:
- A baseline GEO audit — schema, citation graph, AI visibility test queries, content depth
- A 90-day foundation build — schema deployment, citation graph cleanup, About / capability expansion, FAQ hub launch
- Monthly AI citation tracking with reported visibility delta over time
- Ongoing content cadence designed for both traditional search ranking and AI extraction
We're a small new agency. We're not the biggest. We're probably not the cheapest. We are, as far as we can tell, the only agency serving the trades that builds the GEO layer in by default rather than treating it as a future add-on.
If that matters to you, we should talk. If your current setup works and you're happy, ignore this section and just take the playbook above and run it yourself.
7. The Timeline: What to Expect
GEO is not a 30-day play. Anybody who promises you AI Overview citation in your first month is either lucky or lying. Here's the honest timeline.
Months 0–3: Foundation
The technical and structural work. Schema deployment, citation graph cleanup, About / capability content expansion, FAQ hub launch, llms.txt deployment, baseline AI visibility test queries. You will not see citation results yet. You're building the conditions for citation. Have patience.
Months 3–6: Early signals
The first wins start to appear. AI tools begin citing your FAQ content for specific narrow queries. Your business name starts appearing in 1–2 of your test queries (out of 8–12). Google AI Overview begins surfacing your service pages for long-tail queries. You're not winning the headline queries yet. You're laying the citation graph that will eventually win them.
Months 6–12: Citation acceleration
This is where the work compounds. By month nine, you should be cited across multiple AI search surfaces for several of your priority queries. Your competitive moat starts to widen — every new citation, every new piece of completed content, every new month of review velocity reinforces your position. Competitors starting GEO at month nine are not catching up.
Months 12+: The moat
By year two, the GEO foundation built in year one is paying compounding dividends. The companies that started in 2026 will, by 2027, have a citation graph that's effectively impossible to replicate quickly. New entrants to GEO at that point will spend two years getting to where you were after one year — because the algorithms have moved on, the playbook has refined, and your accumulated citations have continued to compound.
This is the part most agencies don't tell their clients. GEO isn't a campaign. It's a moat. The moat takes time to dig. The companies who started digging early get to defend a position the late starters can't reach.
What can go wrong
- Foundation skipped. Trying to do GEO without fixing baseline SEO doesn't work. AI tools pull from the same web infrastructure Google does. A broken foundation breaks both.
- Content that hedges. Tease content doesn't get cited. Complete content does. Most trade businesses fight this because they're afraid of "giving away the answer." Get over it.
- Inconsistent execution. GEO compounds — if you stop posting, stop responding to reviews, stop publishing FAQ content for six months, the citation velocity decays.
- Single-platform fixation. Building only for ChatGPT (or only for Google AI Overview) misses the broader shift. Build for the pattern, not for one tool. The pattern survives the tools.
You've Got the Map. The Window's Open.
Most of your competitors haven't started. Most of the agencies serving the trades aren't even talking about this yet. The window for getting ahead on GEO is genuinely wide open in 2026 — and genuinely closing year over year.
Whether you run this playbook yourself or hand it to us, run it. The trade businesses that get cited by AI in 2026 will compound that visibility through 2027, 2028, and beyond. The ones who wait until "AI search is more proven" will spend years catching up to where the early starters are by then.
If you want help — or if you just want a baseline GEO audit on your business, free — book a 30-minute Growth Audit. We'll pull your current AI visibility, your schema status, your citation graph, and your content depth. We'll show you exactly where you stand and what we'd build in the first 90 days. No pitch deck. No contract.
→ See the dedicated service: GEO & AI SEO for trades → Grade your AI readiness in 3 minutes: Marketing Health Score