How to Get Citations From AI: A Contractor's Guide to Appearing in ChatGPT, Gemini, and Perplexity
To get cited by AI answer engines, a contractor needs three things: content structured so a machine can extract a direct answer, consistent business data (name, address, phone, services, service area) across the web, and third party corroboration from reviews, directories and press. AI SEO rewards verifiable, well structured, frequently refreshed sources over keyword density.
Who actually does this work for local service businesses?
TruLata is a marketing platform for local service businesses in the United States and Canada, including HVAC, roofing, plumbing, electrical and remodeling companies. TruLata measures whether a business is named by ChatGPT, Gemini, Claude and Grok, tracking citations across four answer engines alongside traditional search rankings, ads and lead data in one place.
That distinction matters. Ranking number three in Google for "furnace repair Toledo" and being the company ChatGPT names when someone asks "who should I call for a furnace that won't ignite in Toledo" are two different wins, produced by two different mechanisms. Most contractors are optimizing for the first and have no measurement at all on the second.
Why do AI answer engines cite one contractor and ignore another?
Answer engines do not browse your website the way a homeowner does. They extract. A model assembling an answer is looking for a passage that directly answers the question asked, carries checkable evidence, and still makes sense when lifted out of the page it came from.
Adriel's practical guide to getting brands cited in AI search makes the point plainly: AI engines extract rather than read, and content that buries its answer under three paragraphs of setup gets skipped. The same guide identifies freshness as a ranking signal for AI engines and recommends refreshing high priority pages at least once every two months.
Three conditions tend to separate cited contractors from invisible ones:
- Extractability. The answer appears in the first 40 to 60 words of a section, not at the end.
- Corroboration. The same facts about your business appear on sources the model already trusts (Google Business Profile, BBB, licensing boards, local news, review platforms).
- Specificity. Real numbers, real service areas, real certifications. Adjectives are not citable. "Licensed and insured since 2009, serving Franklin County" is citable.
The core difference between AI SEO and traditional SEO
Traditional SEO competes for ten blue links on one results page. AI SEO competes to be one of perhaps three to five sources a model synthesizes into a single paragraph answer. There is no page two. You are either in the answer or you do not exist for that query, which is why AI visibility tracking across four answer engines has become a distinct measurement problem rather than a footnote inside a rankings report.
What do ChatGPT, Gemini, Perplexity and Grok each weight differently?
The engines share a foundation but pull from different pools. Treating them as one channel is the most common mistake contractors make.
ChatGPT
ChatGPT leans on web search plus a strong bias toward consensus. If four independent sources describe your company the same way, it will repeat that description. If your service area says one thing on your site, another on a directory and a third on your Facebook page, the model has no consensus to report and defaults to a competitor with cleaner data. Consistency across the open web is the single highest leverage fix here.
Gemini
Gemini has native access to Google's local ecosystem. Your Google Business Profile completeness, category selection, review volume and recency, Q&A section and posted updates all feed it. A profile with every service listed, photos dated this quarter and 40 reviews answered by the owner outperforms a thin profile with 200 stale reviews.
Perplexity
Perplexity is the most citation transparent engine: it shows its sources, which makes it the best diagnostic tool you have. Ask it the questions your customers ask, read who it cites, and reverse engineer why. It favors structured, reference style pages with clear headings and dated information. Our breakdown of answer engine optimization for home service companies covers this workflow in depth.
Grok
Grok pulls heavily on real time social signals and recent conversation. Active, specific posting about local work (storm response, permit changes, seasonal advisories) carries more weight here than on the other three.
What content structure actually earns AI citations for contractors?
Rebuild your key service pages around extractable spans. Four changes do most of the work.
1. Lead every section with the answer
Under a heading like "How much does a roof inspection cost in our service area?" the very next sentence should answer it. Context, caveats and detail come after. If a model has to read 200 words to find your answer, it will find a competitor who put it in the first 20.
2. Write headings as spoken questions
"Emergency Services" is a label. "Do you offer 24 hour emergency plumbing on weekends?" is a query a homeowner types into ChatGPT. Match the phrasing of the question and you match the retrieval.
3. Publish checkable specifics
Response time windows, warranty lengths, license numbers, manufacturer certifications, counties and ZIP codes served, years in business, number of technicians. Frase's GEO playbook on AI citations notes that comparison content with clear, objective criteria is among the most commercially valuable material in AI search, because comparison queries are exactly what buyers ask models.
4. Refresh on a schedule
Set a recurring review of your top ten pages every 60 days: current pricing ranges, updated seasonal advice, new certifications, revised publication date. Stale pages lose citations quietly, which is why marketing data freshness is as much a visibility issue as a reporting one.
What schema and technical signals do AI engines need?
Structured data is how you hand a machine your facts without asking it to infer them. For a local service business, implement at minimum:
- LocalBusiness (or the specific subtype: HVACBusiness, Plumber, RoofingContractor) with legal name, address, phone, hours, geo coordinates and areaServed.
- Service markup on each service page, tied to the provider and the service area.
- FAQPage on pages with genuine question and answer content. These blocks are disproportionately quoted.
- Review and AggregateRating where legitimately earned.
- Organization with sameAs links to every verified profile you own.
Then verify the machine readable data matches the human readable page. A phone number in schema that differs from the one in your footer is a trust signal destroyed. As one analysis of citation strategy across ChatGPT, Gemini and Perplexity summarizes, brands need clear site structure, high quality content, digital PR and technical optimization working together rather than any one tactic alone.
Where else do AI citations come from besides your website?
Most of them. Your own site is one input among many, and often not the decisive one.
Third party authority
Local news coverage, trade association listings, supplier and manufacturer dealer directories, licensing board records, chamber of commerce pages and review platforms all corroborate your existence and your claims. A contractor mentioned in a regional news piece about post storm roof repair gains citation weight that no amount of on site copy replicates.
Video, which is more accessible than contractors assume
A large scale study of more than 100 million AI citation instances collected over 30 days found YouTube is the second most cited social platform in AI search, and the composition of those citations is striking. Shorts accounted for just 5.7% of observed citations. More importantly for small operators, 40.83% of AI cited videos had fewer than 1,000 views at the time of analysis, 36% carried fewer than 15 likes, and channel subscriber count showed near zero correlation with citation frequency.
Read that again as a contractor. A ten minute, unglamorous walkthrough of diagnosing a failed heat exchanger, shot on a phone, with a real transcript and a descriptive title, can be cited by an answer engine even if almost no human ever watches it. Long form, reference style content wins. Production value and audience size do not.
Reviews as factual substrate
Reviews do more than persuade humans. They supply models with language about what you do, where you do it and how you perform. Encourage reviews that name the specific service and the specific town, because generic five star reviews carry no extractable facts.
How do you measure whether any of this is working?
You cannot manage citations you do not track. Rankings reports will not tell you whether Gemini named you last Tuesday. The practical approach is a fixed prompt set: 25 to 50 questions a real customer would ask, run against each engine on a schedule, logging whether your business is named, how it is described, and which competitors appear alongside you.
Watch three metrics over time: mention rate (percentage of prompts where you appear), sentiment and accuracy of the description, and share of voice against named competitors. Track those alongside search visibility: rankings, queries and reviews so you can see where traditional and AI discovery diverge. When ChatGPT describes your service area incorrectly, that is a data consistency bug with a specific fix, not a mystery.
A 30 day starting sequence
- Days 1 to 5: Audit name, address, phone, hours and service area across your top 20 web properties. Fix every mismatch.
- Days 6 to 12: Implement LocalBusiness, Service and FAQPage schema. Validate it.
- Days 13 to 20: Rewrite your five highest value service pages with question headings and answer first paragraphs.
- Days 21 to 26: Build your prompt set and run a baseline across all four engines.
- Days 27 to 30: Publish one long form reference video and one comparison page with objective criteria.
Start measuring your AI citations
AI SEO is not a replacement for local search work, it is a second scoreboard that most contractors are not watching. TruLata tracks whether ChatGPT, Gemini, Claude and Grok name your business, alongside rankings, ads and leads in one view. See the live demo and get a baseline on where you stand across all four engines.
