Local SEO blog
Why ChatGPT Names a Directory Instead of Your Contracting Business
When a homeowner asks ChatGPT for a roofer, electrician, plumber, HVAC company, or restoration contractor, the answer may name Angi, Yelp, BBB, or another directory before it names your business.
That feels backwards when your company does the actual work.
I see this problem across local service websites. The contractor has years of experience, strong reviews, and a decent-looking website. The homepage has a huge hero image and a headline about quality workmanship. ChatGPT still pulls its answer from a directory page.
The reason comes down to structure.
Directories are built from clean, standalone facts. Most contractor websites are built around branding, visual design, and broad claims such as “quality you can trust.”
ChatGPT has an easier time extracting a service, a city, a phone number, and a reputation signal from the directory.
The fix is structural. You need to make your own website just as clear and easy to retrieve.
Why the directory wins by default
A directory often wins because it gives an AI system a complete answer in a compact, predictable format.
A typical listing may include:
- Business name
- Trade or category
- Cities served
- Address and phone number
- Hours
- Review count and rating
- Short service description
- Links to the company website
That format answers the basic local question quickly: who provides this service, where do they work, and how can someone contact them?
A contractor homepage often gives ChatGPT more work. The service area may appear in a footer. The actual services may be hidden behind a dropdown. The page may say “complete home solutions” instead of naming roof replacement, emergency plumbing, or electrical panel upgrades.
A beautiful website can still be difficult for an AI system to quote.
This is why local SEO for service businesses requires more than a Google Business Profile and a homepage refresh. Your site needs pages that stand alone as useful answers.
For example, a roofing service page should clearly state that the company provides roof replacement in Grand Rapids, Michigan. An electrical page should state that the company handles electrical panel upgrades in Holland, Michigan.
Service plus city belongs in the same sentence.
I use that test when reviewing contractor pages. If I cannot identify the service and location within a few seconds, an AI crawler may struggle too.
What the Cheers study found
The Cheers home-services study found a close split between contractor websites and third-party directories in ChatGPT citations, while Gemini cited contractor websites much more often.
According to the Cheers home-services AI search study, Cheers ran 32 home-services prompts across 100 US and Canadian metro areas on four engines over the 28 days ending September 2, 2026, and classified 356,019 citations.
Study result: 48.8% of ChatGPT’s home-services citations went to contractor websites. Another 44.7% went to directories and review sites. Gemini cited contractor websites 80.7% of the time.
Those numbers describe citation sources in that study. They do not represent a ranking probability, a guarantee, or a forecast for any specific company.
They do show how important the two information systems are:
- Your own website
- The public listings and review platforms that describe your business
ChatGPT leans on contractor websites for nearly half of its home-services citations in this sample, and it leans on directories and review platforms for the other 44.7%. The gap opens when your site gives the system vague information while the directory hands it a clean answer.
Gemini’s 80.7% contractor-website citation share also gives service businesses a useful reason to improve their owned pages. A strong site supports visibility across multiple engines.
That matters for roofing organic search growth, electrician marketing, and any small business organic growth plan built around high-intent local searches.
How ChatGPT retrieves local business information
ChatGPT appears to retrieve local information through a discovery index, a reading cache, and limited live page opens.
A Search Engine Land report published August 17, 2026 described three layers in ChatGPT’s retrieval process.
The first layer is a discovery index. The report refers to this internal index as “labrador.” A result in that index stores the URL, the full page title, and a snippet of roughly 200 characters.
That snippet matters.
The indexed text is generally anchored on the H1 and includes the visible text directly after it. In some cases, ChatGPT may form an answer from the title and this short snippet without opening the full page.
A directory page often performs well in that environment because its opening content gets to the point quickly:
“ABC Roofing provides roof replacement and storm repair in Grand Rapids, Michigan.”
A contractor homepage may open with:
“Built on integrity. Focused on excellence. Serving West Michigan since 1987.”
That branding has value for a person already comparing companies. It gives an AI system fewer concrete facts to work with.
The second layer is a reading cache. Search Engine Land’s report describes this as a stored full-page copy, often converted into Markdown. ChatGPT can use the cached page when it needs more context.
The third layer consists of limited live page opens. A fresh page fetch can happen, but it does not happen for every answer.
This creates a practical writing rule: put the important answer near the top of every page.
Your H1, opening paragraph, title, and main service description should make the page’s purpose obvious. A contractor should not make ChatGPT dig through three screens of marketing language to discover the service area.
Build service pages that answer one local question
A strong service page answers one clear local question in its opening sentence.
For example:
[Company Name] provides emergency plumbing repair in Kalamazoo, Michigan for burst pipes, sewer backups, and urgent leaks.
That sentence contains the company, service, city, and relevant problem types. It can stand alone in a search result, an AI answer, or a directory-style citation.
The rest of the page can explain process, pricing factors, licensing, project examples, warranties, and frequently asked questions.
I use this structure when working on pages for service businesses because it supports both human readers and retrieval systems. The customer gets an answer immediately. The search engine gets a clear topic. The AI system gets a quotable statement.
This approach fits a no fluff SEO for contractors program. Every section needs a job.
A roofing page should explain roof replacement in a specific service area. An HVAC page should explain emergency heating repair in a specific city. A house restoration SEO strategy should create separate pages for water damage restoration, fire damage restoration, mold remediation, and the locations actually served.
Do not create dozens of city pages with the same paragraph and a swapped city name. Give every important page useful local details, clear service boundaries, and proof that the company works in that market.
You can also read my guides on building topical authority and content SEO strategy for AI search for the larger content structure behind this approach.
Match your schema to the page content
Matching schema helps search systems connect your service page to the business, service, and location described on the page.
A service page about residential electrical work should use schema that supports the actual business and service relationship. The visible page content, business details, service area, and structured data should agree.
At a basic level, review:
- LocalBusiness or a relevant contractor business type
- Service schema for the specific service
- Business name, address, and phone
- Service area details
- Website and profile links
- Opening hours
- Review and rating data, when eligible and accurately represented
Schema cannot turn vague content into a useful answer. It gives machines additional context around clear content.
I treat schema as part of technical SEO, alongside crawlability, internal linking, redirects, and page architecture. The markup should reinforce what the page already says.
Keep your Google Business Profile and entity data current
A maintained Google Business Profile gives local search systems current information about your business, services, locations, hours, photos, and reviews.
Your website and profile should agree on the core business facts.
That includes:
- Business name
- Address
- Phone number
- Hours
- Primary category
- Service categories
- Cities and service areas
- Website URL
This is NAP consistency, but the larger issue is entity consistency. Search systems need to connect the business on your site with the business on Google, Yelp, BBB, trade directories, social profiles, and local publications.
A typo in one directory may seem harmless. Several conflicting phone numbers, addresses, or service areas create a less reliable picture.
The local SEO services page explains the broader work involved in listings, citations, and local visibility.
Directories still matter here. Strong profiles can help ChatGPT discover and confirm your business. Your goal is to make those profiles point back to a site that provides deeper, more useful answers.
Run the free ChatGPT test in your own market
The fastest free test is asking ChatGPT the questions your customers already ask in your city and recording the sources it names.
Start with prompts such as:
- Who are the best roofers for storm damage in Grand Rapids?
- Which electricians install electrical panels in Holland, Michigan?
- Who handles emergency furnace repair in Lansing?
- What should I ask a contractor after water damage in my home?
- Which plumbing companies serve downtown Kalamazoo?
Run the prompts in a fresh chat. Record:
- Businesses named
- Direct website citations
- Directory citations
- Review sites cited
- Pages linked
- Services and locations mentioned
- Missing or incorrect business information
Repeat the test across several days and prompts. One answer is an observation, not a complete measurement.
Pay close attention to the pages ChatGPT cites. If it names your business but links to a directory, your website may need stronger standalone service content. If it never names your business, review your entity data, Google Business Profile, crawl access, local pages, and third-party mentions.
This test works for roofing organic search growth, electrician marketing, HVAC lead generation, plumbing, remodeling, and restoration.
It also gives marketers a practical starting point for small business organic growth plans. You can fix the pages and facts that show up in the actual answers instead of guessing at what an AI system may prefer.
The fix is a clearer information system
ChatGPT names a directory when that directory gives it the cleanest evidence for the question being answered.
You can improve your chances of being named directly by making your website easier to discover, read, understand, and cite.
Put the service and city in the same sentence. Write opening paragraphs that answer the page’s question. Match your schema to the visible content. Maintain your Google Business Profile. Keep NAP and entity data consistent. Build useful service and location pages that help people before they contact you.
That is the practical side of AI search for local contractors.
The work looks a lot like solid SEO because it is solid SEO. The difference is that your pages now need to serve a human reader, a traditional search engine, and an answer engine at the same time.