AI search usually leaves a local contractor out for one of four practical reasons: the system cannot reliably discover the business, cannot match it to the exact service and location, finds conflicting identity information, or lacks enough usable evidence to include it in a short answer. No platform publishes a complete contractor recommendation formula. Google does disclose that its local results rely mainly on relevance, distance, and prominence, while its AI search features use normal Search eligibility and SEO foundations.[1][2]
The fix is not an “AI ranking” trick. It is a clean business record, specific service evidence, credible local proof, crawlable pages, and a lead path that works when someone does click. This article focuses on exclusion points, not the broader contractor AEO operating system.
Platform documentation and public pages were checked September 2, 2026. Recommendation outputs vary by product, query, location, and date.
What the platforms confirm, and what they do not
Google says AI Overviews and AI Mode can run multiple related searches across subtopics and data sources before assembling a response. It calls this “query fan-out.” Google also says a page must be indexed and eligible to appear in Search with a snippet before it can be used as a supporting link. Meeting those conditions does not guarantee that Google will crawl, index, or show the page.[1]
For local results, Google names three main factors: relevance, distance, and prominence. Complete profile information can improve relevance. Distance depends on the searcher’s location or the location Google infers. Prominence can draw on links, reviews, and ratings.[2]
OpenAI documents a separate search crawler called OAI-SearchBot. Sites that opt out will not appear in ChatGPT search answers, except that navigational links may still appear. OpenAI treats this separately from GPTBot, which is used for possible model training.[6]
Those are verified platform facts. The platforms do not publish a universal list of weights for “best HVAC company” or “which roofer should I hire?” They also do not promise that adding schema, collecting a certain number of reviews, or publishing a set number of pages will earn a recommendation.
KKS analysis: a contractor has to pass several practical gates before an answer system can safely name the business. A failure near the top of the chain can make every improvement below it irrelevant.
The six recommendation gates
| Gate | Question the system must resolve | Common contractor failure |
|---|---|---|
| 1. Access | Can the system find and read the relevant page? | Blocked crawler, noindex tag, wrong canonical, server error, or important facts hidden from rendered text |
| 2. Identity | Which real business does this website describe? | Different names, phone numbers, domains, addresses, or unexplained brands |
| 3. Service fit | Does the company perform the exact work in the question? | A generic “full-service contractor” page with no job-specific evidence |
| 4. Geographic fit | Can the company credibly serve this customer? | Huge city lists that conflict with the profile, office location, or operating reality |
| 5. Corroboration | Do outside records support the company’s claims? | The website makes claims that profiles, reviews, licenses, and other public records do not confirm |
| 6. Selection | Is there enough specific evidence to include this company in this answer? | No project detail, weak comparison information, stale proof, or no page that resolves the buyer’s question |
This is an analytical model, not a leaked ranking formula. It gives an owner a useful order of operations. Start with exclusion. Then improve selection evidence.
1. The right page is not eligible or accessible
A contractor can have a polished website and still be absent from an AI answer because the page that proves the service is not usable by the platform. Check the actual service URL, not only the homepage.
- Does the page return HTTP 200 without a login, challenge, or location gate?
- Can search crawlers access it through robots.txt and the hosting or CDN firewall?
- Does the page allow indexing and snippet display?
- Does its canonical tag point to itself rather than another page?
- Can a reader find it through internal links and the XML sitemap?
- Are the important service, location, license, and contact facts present as text?
Google specifically tells site owners to allow crawling, make pages findable through internal links, keep important content in text, and keep Business Profile information current. It says there is no separate AI file or special schema required for AI Overviews or AI Mode.[1]
A blocked page is a technical repair. More content will not solve it.
2. The business identity does not resolve cleanly
Google compiles Business Profile information from official websites, licensed third-party data, user contributions, owner contributions, reviews, photos, and its own interactions with a business.[3] The profile is not an isolated listing. It sits inside a larger record assembled from several sources.
Contractors create identity problems in ordinary ways. A company changes its phone number but leaves the old number on project pages. A DBA appears on the trucks while the website names only the legal entity. A second brand uses the same license without explaining the relationship. A move leaves old addresses in directories and page footers.
Google’s profile guidelines tell businesses to use the name recognized in the real world, keep the address or service area accurate, use the fewest categories needed to describe the core business, and maintain one profile per business.[4]
KKS analysis: minor formatting differences are not the central problem. An unexplained difference in the underlying entity is. Write down the brand, legal entity, domain, main phone, eligible address or service-area status, license holder, and primary trade. Then compare that record with the website, Business Profile, state license record, major directories, and association listings.
3. The website names a trade but not the requested service
“Plumber” does not answer whether a company replaces sewer lines, repairs slab leaks, installs tankless water heaters, or takes emergency calls after 8 p.m. “Remodeler” does not establish experience with accessible bathrooms, structural additions, or occupied-home kitchen projects.
Recommendation questions are often narrower than category pages. Google says AI Mode can handle nuanced questions and complex comparisons by searching related subtopics.[1] A contractor site needs pages that resolve those subtopics with facts a homeowner can use.
A useful service page should state what work is included, what is excluded, who the service fits, the normal process, relevant equipment or materials, geographic limits, and the next step. Add original project evidence when available. Do not invent local work to fill a template.
The broader SEO guide for contractors explains where service and market pages belong. The rule here is narrower: if the requested job cannot be confirmed from one or two clear pages, the business has a service-match problem.
4. The geography is asserted rather than demonstrated
Distance is a structural limit in Google’s local system. A page cannot move a shop closer to the searcher.[2] A contractor can improve how clearly the site and profile describe the real operating area, but a hundred city names do not cancel geography.
Check whether the company’s stated coverage agrees across the Business Profile, contact page, service pages, dispatch policy, and project history. For a service-area business, do not publish a customer-facing address that is not eligible. Do not claim a market because one crew once drove through it.
KKS analysis: a strong local proof set combines a credible service radius, completed work in the market, city-specific constraints when they matter, and a conversion path that can actually serve the area. The site should make unprofitable or unsupported territory less prominent, not advertise every city within driving distance.
5. The company’s own claims stand alone
A website can state that a company is licensed, experienced, responsive, and trusted. An answer system still has to decide whether those claims deserve use. Google says its local prominence factor can draw on links and reviews, and that more reviews and positive ratings can help local ranking.[2]
That does not mean review count controls every AI answer. It means independent evidence matters in at least one documented local search system. Contractor owners should inspect the evidence attached to each important claim:
- Does the state record match the business or qualifying party named on the site?
- Do reviews mention the service being targeted, without coaching customers to use scripted language?
- Do manufacturer or trade-association listings point to the right domain and phone?
- Do local project pages show work the company can substantiate?
- Do old directories create a second phone number, address, or business name?
A public license is proof of a credential only within the issuing jurisdiction and class. It is not proof of workmanship, customer satisfaction, or current capacity. Keep those distinctions intact.
6. The page is hard to quote or compare
Some contractor pages contain plenty of words but little usable information. “Quality service you can trust” cannot answer a question about permit handling, emergency availability, equipment brands, warranties, financing, or project type.
Make the main answer visible early. Use descriptive headings. Put differences into a table when a table helps the buyer. Define terms. Show dates on facts that can go stale. Link to the source of a license, certification, policy, or program when the public record is available.
This does not require writing for robots. Google’s published guidance says the same people-first content and normal SEO foundations apply to its AI features.[1] The practical test is whether a cautious homeowner could understand and verify the page.
7. Schema is being asked to prove what the page does not
LocalBusiness structured data gives a site a standard way to state business details. Google documents fields for items such as business type, address, hours, telephone, and URL.[5] Schema can reduce ambiguity when it matches the visible page.
It cannot create a license, verify a service area, repair a stale Business Profile, or make a thin page useful. Google tells site owners that structured data should match visible text and that no special schema is needed for its AI features.[1]
Use schema after the underlying facts are settled. Validate the code, inspect the rendered page, and check the exact URLs. If the markup and visible page disagree, fix the record rather than choosing whichever version looks better.
A 30-minute recommendation exclusion audit
Run this in order on one profitable service and one real market. A narrow test exposes more than a vague search for the company name.
- Write the buyer question. Example: “Who installs cold-climate heat pumps in [city]?”
- Name the proof page. Identify the one URL that should establish service fit.
- Test access. Confirm HTTP 200, indexability, snippet eligibility, canonical, rendered text, internal links, and crawler rules.
- Compare identity fields. Check brand, domain, phone, address or service-area status, and license relationship across the site and major records.
- Check service specificity. Look for scope, exclusions, process, equipment or materials, proof, and a working next step.
- Check geographic truth. Confirm the company serves the location profitably and has evidence that supports the claim.
- Check outside corroboration. Review current public records and profiles for contradictions.
- Record the result. Save the exact query, product, date, location context, companies named, and cited URLs. Repeat the same test later before claiming movement.
Do not grade the company from one answer. Generated results can vary. The audit should find controllable defects in the evidence chain, not manufacture a score from a screenshot.
Find the exclusion point before buying more content
The free KKS contractor website and competitor audit reviews 15 factors and up to three relevant local competitors. When the evidence supports it, the audit includes search, demand, license, and ownership findings, plus a 90-day action plan. Qualified requests are delivered within two business days.
Repair, rebuild, or leave it alone?
Repair is usually enough when the site is owned by the contractor, important pages are editable, technical access works, and the problem is limited to identity fields, profile alignment, internal links, service detail, or conversion paths.
A controlled rebuild deserves consideration when the contractor cannot obtain administrator access, the platform blocks necessary changes, URL and template problems affect most of the site, or the current vendor cannot transfer the assets needed for a safe exit. Review contracts, analytics ownership, redirects, forms, call tracking, and backups before moving anything. The contractor marketing agency buyer’s guide has a shared control and exit checklist for that decision.
No engagement is the honest answer when the website is not the binding constraint. If the company cannot take more work, does not answer leads, lacks authority for the advertised service, or has an unresolved ownership dispute, fix that issue first.
The broader contractor search strategy connects this evidence work to Maps, organic search, reviews, AI answers, and booked-job measurement. The exclusion audit has one job: find the earliest broken gate so the next dollar goes to the right repair.
Frequently asked questions
Can I pay to make AI recommend my contracting company?
No major platform cited here offers a paid guarantee of an organic AI recommendation. Google says there is no way to request or pay for better local ranking. Advertising is a separate product from organic local or AI inclusion.[2]
Does ChatGPT use my Google Business Profile?
OpenAI’s crawler documentation does not identify Google Business Profile as a universal source for local recommendations. Treat claims about a fixed ChatGPT source mix with caution. OpenAI does confirm that OAI-SearchBot is used to surface websites in ChatGPT search results.[6]
Will LocalBusiness schema make AI recommend my company?
No. Schema can state business details in a machine-readable format, but Google says no special schema is required for its AI features. The markup should agree with the facts visible on the page.[1][5]
How many reviews do I need before AI recommends my business?
No verified universal threshold exists. Google says review quantity and positive ratings can help local prominence, but it does not publish a number that guarantees a local result or AI recommendation.[2]
How often should I test AI visibility?
Use a fixed set of commercially relevant questions and retest monthly after material changes. Record the platform, exact query, date, location context, answer, and cited pages. One favorable response is a signal, not a stable ranking.
Sources
- AI features and your website | Google Search Central. Accessed September 2, 2026.
- Tips to improve your local ranking on Google | Google Business Profile Help. Accessed September 2, 2026.
- How Google sources and uses Business Profile information. Accessed September 2, 2026.
- Guidelines for representing your business on Google. Accessed September 2, 2026.
- Local Business structured data | Google Search Central. Accessed September 2, 2026.
- Overview of OpenAI crawlers. Accessed September 2, 2026.