AI Referral Checklist: Get Your Home Service Business Recommended by ChatGPT & AI
Co-Founder & CRO at Driive
Nick Small is the Co-Founder and Chief Revenue Officer at Driive, where he helps home service contractors modernize their booking and scheduling operations. Before Driive, Nick spent a decade in B2B SaaS helping field service businesses grow.
Learn more about our team28
Checklist items to complete
3
Pillars: Specificity, Consistency, Proof
100%
Free to implement yourself
To get recommended by AI assistants: (1) be specific — name your exact services and service areas on your website, (2) be consistent — same business name, phone, and address across every directory, (3) build proof — reviews on multiple platforms and third-party citations, and (4) create AI-readable pages — an llms.txt file and an /about-for-ai page. This checklist walks you through all 28 steps.
Homeowners used to ask their neighbor. Then they Googled. Now a meaningful chunk are asking ChatGPT, Gemini, and Perplexity who to call. The contractors getting recommended are not the ones with the biggest ad budgets. They are the ones the models can understand.
This checklist is the operator version of the playbook we covered in our webinar with CompanyCam. Watch the recording or register for upcoming sessions here.
Go through the checklist below and check off everything you already do. The items you can't check? That's your to-do list.
The Three Pillars
Every contractor we talk to asks the same question: how do I get ChatGPT to recommend my business? The answer comes down to three pillars. Miss any one of them and the models will recommend someone else.
Be specific
Your website highlights what's unique about your business.
Be consistent
One business name, one phone, one address format. Everywhere.
Have proof
Across multiple platforms, not just Google.
Section A. Specificity (your website)
AI models do not read between the lines. They read lines. If your website says "trusted local pro" without naming which services you offer, which cities you serve, and what specific problems you solve, the models have nothing to work with.
The test is simple: if you swapped your company name for a competitor's name on your homepage, would a reader notice? If the answer is no, your website is too generic for AI to recommend.
What specificity looks like in practice:
Bad: "We provide quality plumbing services to the greater metro area."
Good: "Same-day water heater replacement in Douglas County. We service Omaha, Papillion, La Vista, and Bellevue. Licensed, bonded, and available 24/7 for emergencies."
The second version names the service (water heater replacement), the geography (Douglas County plus four cities), and a differentiator (same-day, 24/7). That is what the models can use.
Section B. Consistency (your listings)
AI models cross-reference. They look at your website, your Google Business Profile, Yelp, BBB, Angi, Nextdoor, and dozens of other directories. If the information does not match, the models lose confidence. Worse, they may recommend a competitor whose information is cleaner.
Consistency is not just about having the same information everywhere. It is about having identical information everywhere. Same capitalization. Same punctuation. Same abbreviations. Same phone format.
Examples of consistency failures:
Business name: "Smith & Sons HVAC" on your website but "Smith and Sons Heating" on Yelp and "SMITH AND SONS" on BBB. The AI sees three different businesses.
Phone number: "(402) 555-1234" on Google but "402.555.1234" on your website and "402-555-1234" on Facebook. Small differences, but they add up.
Address: "123 Main Street, Suite 200" versus "123 Main St Ste 200" versus "123 Main Street #200". Pick one format and use it everywhere.
What directories should you check? Start with the top 10: Google Business Profile, Yelp, BBB, Angi, Nextdoor, HomeAdvisor, Thumbtack, Facebook, Apple Maps, and Bing Places. Then expand to the next 20: Yellow Pages, Manta, Mapquest, CitySearch, Superpages, your local chamber of commerce, trade association directories, and supplier directories. Most contractors have listings they forgot they created. Find them and fix them.
Section C. Proof (beyond Google)
Google reviews are table stakes. Every contractor has them. What separates the businesses that get AI recommendations from the ones that do not is proof that exists outside the Google ecosystem.
AI models are trained to look for third-party verification. Reviews on multiple platforms. Mentions in forums. Citations in local news. Before-and-after photos tied to specific jobs. The more independent sources confirm your business, the more confident the model is in recommending you.
The Reddit and Quora test: Search your business name on Reddit and Quora. What comes up? If the answer is nothing, you have a proof gap. If someone asked "who should I call for HVAC in [your city]" and you are not in the thread, you are invisible to that entire conversation. And AI models read those threads.
Why specific review language matters: A review that says "Great company, highly recommend" tells the model nothing. A review that says "Called Smith & Sons for a furnace replacement in Papillion, they showed up same day and had it installed by evening" tells the model exactly what you do, where you do it, and how fast. Encourage customers to be specific. It helps you rank in traditional search and it helps AI models recommend you.
Section D. LLM-readable pages
Here is something most contractors do not know: you can create pages specifically for AI models to read. Not hidden pages. Not keyword stuffing. Clean, structured pages that summarize exactly who you are, what you do, and where you do it.
The llms.txt file: This is a plain text file that lives at the root of your website (yourdomain.com/llms.txt). It points AI models to your most important pages. Think of it like a table of contents for robots. We include a template below.
The About-for-AI page: This is a page on your website written in plain language specifically for AI agents to read. It covers the basics: who you are in one sentence, services by name, cities and zip codes served, hours, response time, payment options, and what you do not do. The "what you do not do" part is important. It helps AI models avoid recommending you for jobs you cannot take.
Section E. Test it
You will not know if any of this is working unless you test it. And the test is simple: ask the AI models yourself.
Open ChatGPT, Gemini, and Perplexity. Ask: "Who should I call for [your trade] in [your city]?" Do you come up by name? If yes, you are ahead of most contractors. If no, you now know which pillar to focus on.
Run this test monthly. AI models update their training data periodically. A recommendation that works today may not work next month if a competitor improves their presence. Treat this like checking your Google reviews. It takes two minutes and tells you where you stand.
The Checklist
Below is the full 28-item checklist across all five sections. Click to check off what you already do. Your progress is saved automatically.
Template: your llms.txt file
Copy this template and save it as llms.txt at the root of your website. Replace the bracketed content with your actual business information.
llms.txt is a plain text file that lives at the root of your website. It points AI models to your most important content. Copy the prompt below into ChatGPT or Claude, fill in your business details, and send the output to your developer.
I run a [trade, e.g. HVAC] business based in [city, state]. My business name is [exact legal name]. My website is [domain]. I serve the following cities and zip codes: [list them]. My main services are: - [service 1 with short description] - [service 2 with short description] - [service 3 with short description] My most important pages are: - [URL]: [what it covers] - [URL]: [what it covers] - [URL]: [what it covers] Please write me a complete llms.txt file following the llms.txt specification. Start with an H1 of my business name. Add a one-line summary as a blockquote. Group my pages into sections by topic (services, about, locations, contact). Format each link as: - [Title](URL): one-line description. Do not invent pages or services I did not list. Output only the llms.txt content, nothing else.
Here is an example of what the output looks like:
# Sarpy HVAC > Same-day AC and furnace repair in Sarpy County, Nebraska. ## Services - [AC Repair](https://example.com/services/ac-repair): same-day diagnostics and repair for residential AC systems - [Furnace Install](https://example.com/services/furnace-install): new install and replacement on Trane and Carrier ## Service Areas - [Bellevue](https://example.com/locations/bellevue): same-day service across 68005, 68123, 68147 - [Papillion](https://example.com/locations/papillion): same-day service across 68046, 68133 ## Contact - [Book Now](https://example.com/book): real-time booking - [Call Us](https://example.com/contact): 24/7 phone support
Save the output as llms.txt. Upload it to the root of your domain. Confirm it loads at yourdomain.com/llms.txt.
Template: your About-for-AI page
Create a new page on your website at /about-for-ai or /llm-info. Use this template as a starting point. The goal is a clean, structured summary that AI agents can read and understand in seconds.
Most contractor websites are written for humans. One page should be written for the other audience. Copy the prompt below into ChatGPT or Claude, fill in your details, and send the output to your developer to publish at /llm-info or /about-for-ai. Driive's own version lives at getdriive.com/llm-info.
I run a [trade, e.g. HVAC] business based in [city, state]. My business name is [exact legal name, as it appears on my Google Business Profile]. My website is [domain]. My phone number is [number]. My address is [address]. Please write me a plain-language web page intended to be read by AI agents (ChatGPT, Gemini, Perplexity, Claude) so they can accurately recommend my business when homeowners ask for help. Cover, in this order: 1. Who we are. One sentence. Include the legal business name, the trade, and the primary service area. 2. What we do. List every service we offer by name. For each service, give a one-sentence description of what it includes and who it is for. Do not invent services I did not provide. 3. Where we work. List every city, county, and zip code we serve. Be specific. 4. When we are available. Hours of operation. Typical response time. Whether we offer same-day, next-day, or emergency service. 5. How to book us. Phone number. Booking URL. What information a homeowner should have ready. 6. What we do not do. Just as important. List services we are commonly asked for but do not offer (example: "we do not do commercial HVAC, only residential"). 7. Proof. Link to our Google Business Profile, our top three review platforms, and any third-party recognition (local news, trade publications, awards). Write in plain language. No marketing fluff. No buzzwords. No em dashes. Short, declarative sentences. Output only the page content, no preamble.
Publish the output as a standalone page. Link to it from your sitemap and from your llms.txt. Driive's version is at getdriive.com/llm-info if you want to see what it looks like in the wild.
The bigger frame
AI is not replacing the trades. It cannot tear off a roof. It cannot come to your house and figure out your toilet flapper is shot. The robots that have tried physical work do not do it well.
What it can do is move transactional work faster. Getting a homeowner from "I need help" to a booked job on your calendar. That is the leverage. The contractors who win the next decade will not be the ones who automate the most. They will be the ones who get recommended the most, and who put rails on the parts of the business they automate.
Frequently Asked Questions
There is no guaranteed timeline. AI models like ChatGPT update their training data periodically, and results depend on how consistently your business information appears across the web. Focus on the fundamentals: be specific, be consistent, and build proof. Most contractors see changes within 3-6 months of implementing these practices.
Ready to transform your booking?
Join hundreds of field service teams already using Driive to win more jobs, reduce drive time, and automate their booking.
