TL;DR: Service businesses are the least exposed group in AI search, and most of the advice sold to them is written for publishers. AI Overviews appear on 36% of informational queries and on 8% of commercial ones, which is where contractors, clinics, and professional firms actually sell. The work that matters is vertical-specific and it is not schema.
The one-sentence answer: For a service business the AI search problem is not lost blog clicks, it is whether an assistant can name you, describe what you do accurately, and hand a ready-to-buy person your phone number, which depends on entity consistency and published operational detail rather than on any technical markup.
A roofing client asked me last year whether he needed to rewrite his site for AI. He had read that AI would take his traffic.
His traffic was fine. His problem was that ChatGPT could not tell anyone what counties he served.
That gap is the real one for service businesses, and it is a different problem from the one publishers have. The general theory is in our generative engine optimization guide and the local mechanics are in AI SEO for local businesses. This post is about what differs by vertical, because a contractor, a clinic, and an accounting firm face three genuinely different versions of this.
Why are service businesses less exposed than publishers?
Because AI Overviews follow query intent, and service businesses live on the intents where they appear least.
Seer Interactive's February 2026 data, drawn from 5.47 million queries and 2.43 billion organic impressions across 53 brands, splits it cleanly: 36% of informational queries returned an AI Overview, against 8% of commercial queries and 5% of transactional ones (Seer Interactive, April 24, 2026).
A publisher's entire inventory is informational. A contractor's revenue sits on commercial and local terms. Same feature, very different exposure. When a national outlet reports a traffic collapse, it is reporting from the worst-hit position in the market, and a service business reading that as its own forecast is reading someone else's weather.
The loss that is real for you is narrower: the top-of-funnel explainer content. Pew Research Center found users clicked a traditional result in 8% of visits where an AI summary appeared, against 15% without (Pew, July 22, 2025). Your "how much does a roof coating cost" post is exposed. Your service-area page is not.
What actually changes for a contractor?
The assistant needs to be able to answer three questions about you, and most contractor sites answer none of them in a form a machine can extract: where do you work, what exactly do you do, and who is it for.
Service area is the one that fails most often. A page that says "serving the greater metro area and surrounding communities" is unusable. A page that names counties and towns is quotable, which is the extraction problem covered in our answer engine optimization explainer. This is not an AI trick, it is the same specificity a customer wants, written where a machine can find it.
The second gap is job-level detail. Contractors know an enormous amount that never gets published: which substrate fails first, what a real square-foot range looks like by system, what a job costs to do properly versus what the low bid omits. That material cannot be summarized away, because until you publish it the assistant has nothing to summarize. Definition posts get answered for free. Judgment does not.
Then there is a feature most contractors have not registered. On July 16, 2025, Google began rolling out a Search feature that calls businesses on a user's behalf to check pricing and availability, across Search users in the U.S. with higher usage limits for AI Pro and AI Ultra subscribers. Businesses control participation through Business Profile settings. Someone searching "roofers near me" may reach you as an automated call asking about availability before any human dials. Whoever answers the phone at your shop is now part of your search presence.
Field-service operators can go further, because operational data is the moat here: real job durations, real crew counts, real scheduling constraints. Everything we build into My ClickWerxs produces that record as a side effect of running the work, which is the cheapest content research a trades business will ever get.
What changes for a healthcare practice?
Healthcare sits at the hard end. Medical questions are overwhelmingly informational, which is exactly the intent class where AI Overviews appear most, and clinical content carries constraints that marketing content does not.
The practical split is between the symptom content and the practice content, and they should be treated as separate programs.
Symptom and condition pages compete directly against a summarized answer, and against institutions with far more topical authority than any local practice will accumulate. Publishing another "signs of X" page spends budget on the one contest you are structurally set up to lose.
Practice-level content has no such competition. Which insurance you accept. Whether you take new patients this month. What the first visit involves, how long it runs, what to bring. Parking. Language coverage. Whether a referral is required. None of that exists anywhere else, all of it is what a patient actually asks an assistant, and none of it is clinical claim-making. It is also the material a front desk answers by phone all day, which is the angle that tends to get a practice manager's attention.
The compliance perimeter is real and it does not change here. Anything that constitutes medical advice, patient testimonials, or outcome claims goes through the same review it always did. AI search does not create a new exemption, and a practice that starts publishing clinical assertions to chase citations has traded a marketing problem for a regulatory one.
What changes for an accountant or a law firm?
Professional services face the version of this nobody wants to say out loud: the assistant is not just summarizing your content, it is doing a piece of your work.
"How do I calculate depreciation on a vehicle" was a lead magnet. It is now a free, correct, instant answer. That category of content is gone as a traffic strategy and pretending otherwise wastes a budget.
What survives is what carries liability. An AI answer cannot sign a return, represent you in an audit, or accept responsibility for being wrong. So the content that still works is the content sitting next to that line: where the general answer breaks down, what the exceptions cost, which situations need a person and why. "When the standard advice is wrong for your situation" is a post an assistant cannot preempt, because the whole point is the judgment call.
The second move is jurisdiction. General tax and legal answers are national and commoditized. State-specific, county-specific, industry-specific answers are narrow enough that no general model has been trained on a confident version, and specific enough that the person searching is already qualified.
What does Google say you actually need to do?
Less than the vendors selling AI SEO packages are telling you, and this is worth checking against your last proposal.
Google's guidance, last updated July 10, 2026, states that structured data is not required for generative AI search and there is no special schema. The exception worth watching is Google's Open Knowledge Format.org markup to add. It states that machine-readable AI text files, llms.txt among them, are not used by Google Search. It states there is no requirement to break content into small pieces, and no need to write in a particular style for generative AI. The stated eligibility bar is that a page is indexed and eligible to be shown with a snippet.
It also names the local exception directly: Google Business Profile helps products and services appear in AI responses as well as in ordinary Search results. For a service business that single line outranks most of the technical advice, because your Business Profile is doing more of this work than your website markup is.
None of that means structured data is useless. It means it is not the lever, and if an agency's AI SEO proposal is mostly markup and an llms.txt file, you are buying activity.
What should every service business do first?
Four things, in this order, before any content strategy.
Check that you are not blocking the retrieval crawlers. Open your robots.txt. Each vendor runs a training crawler and a separate search crawler, and only the second decides whether you can be cited. OAI-SearchBot surfaces sites in ChatGPT's search features, PerplexityBot surfaces and links sites in Perplexity results, and Claude-SearchBot improves search result quality; GPTBot and ClaudeBot are the training crawlers, and blocking those costs you nothing in citations. Google-Extended is a separate case again, since Google states it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search." In our site audits a blanket block from a security plugin or CDN toggle is the single most common silent cause of AI invisibility, and it is almost always a default rather than a decision anyone made. The crawler-by-crawler walkthrough is Check 2 of our AI SEO audit checklist.
Make your entity consistent. Name, address, phone, and service area identical across your site, Business Profile, and major directories. Assistants assemble an entity from scattered mentions, and contradictions produce hedged answers or no answer.
Publish your operational specifics. Service area by named place. Hours including the real emergency policy. What you do not do, which is more useful than it sounds and disqualifies bad leads for free.
Ask the assistants about yourself. Query ChatGPT, Perplexity, and Gemini for your business by name, then for your category plus your city. Record what comes back. The errors you find in twenty minutes are usually the whole first quarter of work.
What happens after the assistant sends someone to you?
This is the step that quietly wastes the work, and it sits outside SEO entirely.
Someone who arrives after an assistant described your business is further along than an ordinary search visitor. They have already been told what you do and that you are a reasonable option. They are closer to booking and less tolerant of a slow reply.
The pattern we see repeatedly in CRM onboardings is the same in every vertical: the largest recoverable group in an old system is not cold contacts, it is warm leads who submitted a form or answered an ad and then heard nothing. Automated follow-up sequences consistently recover those at higher rates than any manual process reached. A business that earns AI citations and then answers a form fill in four days has bought better traffic for the same leak.
Two fixes, neither of them content work. Put a "how did you hear about us" field on your intake form, because that is the only instrument that catches influence without a click. Then make sure something automatic replies within minutes, whether that is ClickWerxs Command Center or anything else that actually fires.
Where does this advice fail?
Three cases, and they are worth checking against your own situation before you act on any of the above.
If a service business is genuinely national and informational at the point of sale, an online-only insurance broker for instance, the publisher's problem is your problem and the intent split above does not protect you.
If your category has heavy franchise or aggregator presence, home services marketplaces most of all, the assistant may answer with the aggregator instead of any individual operator. Entity work still helps, but a directory outranking you inside an AI answer is a distribution problem, not a content one.
And the intent figures here are a February 2026 snapshot from one dataset. Trigger rates have moved every quarter since AI Overviews launched. Run your own twenty queries rather than trusting a percentage, including this one.
Frequently Asked Questions
Do I need a separate AI SEO budget, or is this the same work?
Mostly the same work with a different brief. The technical foundation is ordinary SEO, and Google states no special optimization is required. What changes is what you commission: fewer definition posts, more operational specifics and judgment content, plus citation measurement you probably are not running yet.
My competitor gets mentioned by ChatGPT and I do not. How do I fix that?
Start by checking whether you are crawlable at all, then compare entity consistency across both businesses. A competitor with a complete Business Profile, consistent details across directories, and specific service-area pages gives an assistant enough to answer with, while a vaguer profile gives it a reason to skip you.
Is this worth doing for a business that gets most work from referrals?
It is worth the four fixes above, which take an afternoon. A referred customer still searches your name before calling, and an assistant that cannot describe you accurately costs you a warm lead rather than a cold one.
How do I know if any of it is working?
Measure citations rather than rankings, which is a separate setup and covered in tracking whether AI systems cite your business. Add a "how did you hear about us" field to your intake form in the same week, because the influence you most want here arrives without a click.
The reassuring part is that almost none of this is new work for a business that already documents what it does carefully. The uncomfortable part is that vagueness used to be survivable, and it is not any more.
If you want your entity, your service-area content, and your citation tracking built together, see ClickWerxs AI SEO or get in touch.
Sources
- Seer Interactive, "AIO Impact on Google CTR: 2026 Update" — 36% of informational queries returned an AI Overview; 8% of commercial and 5% of transactional queries; 53 brands, 5.47M queries, 2.43B organic impressions, January 2025-February 2026. seerinteractive.com (April 24, 2026; retrieved 2026-08-22)
- Pew Research Center, "Google users are less likely to click on links when an AI summary appears in the results" — 8% click rate with an AI summary vs 15% without; 900 U.S. adults, 68,879 searches, browsing data March 1-31, 2025. pewresearch.org (July 22, 2025; retrieved 2026-08-22)
- Google Search Central, "Google's Guide to Optimizing for Generative AI Features on Google Search" — structured data not required and no special schema.org markup needed; AI text files including llms.txt not used by Google Search; no content-chunking requirement; eligibility requires indexing and snippet eligibility; Business Profile helps services appear in AI responses. developers.google.com (last updated July 10, 2026; retrieved 2026-08-22)
- Google, "New AI features in Google Search: Call a business or do research" — Search can call businesses for pricing and availability; "starting to roll out to all Search users in the U.S." with "higher limits for Google AI Pro and AI Ultra subscribers"; "businesses are always in control via their Business Profile settings." blog.google (July 16, 2025; retrieved 2026-08-22)
- OpenAI, "Bots" —
OAI-SearchBotsurfaces websites in ChatGPT's search features;GPTBotcrawls content for model training. developers.openai.com (retrieved 2026-08-22) - Perplexity, "PerplexityBot" — surfaces and links websites in Perplexity search results; not used for training. docs.perplexity.ai (retrieved 2026-08-22)
- Anthropic, "Does Anthropic crawl data from the web?" —
Claude-SearchBotfor search result quality;ClaudeBotfor training. support.claude.com (retrieved 2026-08-22) - Google Search Central, "Google crawlers" — "Google-Extended does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search." developers.google.com (retrieved 2026-08-22)
- ClickWerxs site-audit pattern — CDN and security-plugin AI-crawler blocking as the most common silent cause of AI invisibility found in audits. Operator data, on record.
ClickWerxs sells AI SEO services built on the methodology described in this post. This post is general marketing guidance, not legal, medical, or financial advice. Healthcare and professional-services firms remain responsible for their own regulatory and licensing review of published content. Third-party figures reflect the publication dates cited and change frequently.
Kaleb Dickhaut — Founder, ClickWerxs. Kaleb built ClickWerxs from the ground up, from payment processing ISO to the Command Center platform to the AI SEO methodology the blog runs on. He has onboarded hundreds of small businesses onto payment and CRM systems. linkedin.com/in/kaleb-dickhaut