How customers find a dentist, lawyer, or CPA through AI search.
Restaurants, plumbers, and inns are easy to talk about because the customer journey is short. A professional services purchase is different. A new patient picking a dentist, a small business owner choosing an attorney, a homeowner hiring a CPA, an adult booking a first injectable appointment. These customers research before they ever pick up the phone. In 2026, that research increasingly happens inside an AI assistant. Here is what we have seen across the professional services side of our Portsmouth NH audit work, and why the gap is more expensive here than anywhere else.
Why professional services are the harder problem
Two things make professional services different from the consumer categories we have written about elsewhere.
The customer is doing real homework. Someone choosing a dentist for their family is not pattern-matching off a search ad. They are asking ChatGPT or Gemini something specific: who handles anxious patients well, who takes our insurance, who has both general and pediatric, who is actually accepting new patients. The AI does not just retrieve a list. It synthesizes an answer. The businesses that get named in that synthesis win the consideration set. The businesses that do not, never enter the room.
The lifetime value of one new customer is enormous. A new dental patient is worth thousands of dollars over a few years. A small business legal client is worth tens of thousands over a relationship. A CPA client is worth annual recurring revenue for as long as the practice exists. A med spa member is on a monthly cadence. Missing one referral conversation a week is not a marketing inconvenience. It is real money compounding into a serious leak.
In the professional services audits we have run in Portsmouth, the median number of AI engines that retrieve the business on a direct branded query is 4 of 8. On a category query without the brand name (the way a new customer actually asks), the median drops to 1 of 8. The site is fine on Google. It is invisible to the new customer.
The four patterns we see, by vertical
Dental
The hardest fight in dental is the chain practices. Aspen Dental and the corporate-backed groups have enormous structured content, consistent listings, and full schema. They show up first in most AI answers about Portsmouth dentists, often above the family practices that have served the same patients for thirty years. The local practices are not absent on Google. They are absent from the AI summary that increasingly precedes Google.
Specific patterns we see on owner-operated dental sites in the Seacoast:
- The practice has multiple dentists, but the site reads as if it has one. Each provider needs their own crawlable bio, not a single team photo.
- Services exist as a navigation item but are not described in plain language a model can quote back. "General dentistry" is not enough. "Cleanings, exams, fillings, crowns, root canals, extractions" is.
- "Accepting new patients" is the most-asked question and the least-answered one on most sites. If you accept new patients, say so in crawlable text on the homepage.
- Insurance accepted is often listed as logos, which are invisible to most retrieval. The model needs the names spelled out.
Legal
The professional services vertical with the largest gap between human reputation and AI retrieval. A Portsmouth attorney can be the most-referred small business lawyer in the Seacoast and still not appear when ChatGPT is asked "who should I talk to about forming an LLC in New Hampshire." Word of mouth does not propagate through AI engines. Structured signals do.
Specific patterns we see on small and mid-sized law firm sites:
- Practice area pages exist but read as menus, not as content. A model retrieving "small business attorney Portsmouth NH" rewards pages that actually describe the work, not pages that list it.
- Attorney bios are long-form and rich, but the site does not tie each attorney to the practice areas they handle in a way a model can navigate.
- The firm's geographic service area (NH, ME, MA) is implied through the address but never declared in crawlable text. So the engine guesses.
- Long-established firms (1908, 1989, 1980) bury the trust signal in an About page. The founding year belongs on every page that matters.
Accounting and CPA
The CPA vertical is interesting because it is the one where local search still works reasonably well, but AI search performs worst for the firms with multi-location footprints. A firm with Dover, Rochester, and Portsmouth offices often shows up under one location and never the other two. The model picks the office it has the most signals for and treats the others as ghosts.
Specific patterns we see on CPA firm sites:
- Multi-office structures need per-office pages with real content, not a contact-page list. Otherwise only the headquarters office is retrieved.
- Specialty services (cost segregation, business valuation, non-profit work) are gold for AI retrieval because they are specific queries with low volume, but most firms bury them under "services."
- Tax season seasonality is invisible. A firm accepting new clients in March is treated by the model the same as one not accepting until July.
- Founder names anchor trust. Sites that lead with the founder by name retrieve better than sites that lead with the firm name alone.
Medical aesthetics and wellness
Med spa is the vertical where booking behavior has already moved into AI search the most. Adults considering a first injectable appointment, a laser treatment, or a non-invasive procedure ask the assistant before they ask a friend. They want to compare without feeling judged. The assistant happily compares. It compares with whatever signals it has, which means the practices that are loud about their providers, their treatment menu, and their consultation process win.
Specific patterns we see on boutique med spa sites:
- Founder and provider credentials are the most retrieval-valuable content on the site. RNs, NPs, MDs, years of experience, and the medical director relationship matter to the model in a way they do not on a restaurant site.
- Treatment menus are often image-heavy. The model cannot read the images. The same menu in crawlable text retrieves cleanly.
- Membership programs and financing options are common questions and rarely findable on-page in a model-readable form.
- "What is this treatment" content is the highest-converting category in this vertical and the easiest to write. Most boutique spas leave it to the manufacturer's site.
What we recommend, in this order
For an owner-operated professional services business in Portsmouth NH, the work below is the part most owners can think about themselves before they pay anyone to do it. We will say plainly when we think a step is worth hiring out.
- Run the eight tests yourself. Open ChatGPT, Gemini, Claude, Perplexity, Copilot, Meta AI, Grok, and a Google search that triggers AI Overviews. Ask each one the question your best client would ask if they did not already know you. Do not name yourself. Read the answers. Note the gap between the eight.
- Fix the entity statement. Make sure your homepage has a single, crawlable sentence that declares what kind of business you are, who you serve, and where. Most sites have this in marketing copy and not in indexable text.
- Fix the provider, attorney, or principal pages. Each professional gets their own URL, their own bio, their own crawlable list of what they handle, with the credentials and tenure stated plainly.
- Audit the schema. The structured data on most professional services sites is wrong, stale, or missing. This is the single highest-leverage technical fix and the one we usually hear "I had no idea" about. We will say where it is worth hiring this out.
- Reconcile the listings. The model averages across every place your business appears: Google Business, the bar association directory, the state CPA society, the chamber of commerce, the insurance network listings. One inconsistency confuses the model. Five confuse it badly.
- Write the missing pages. Most professional services sites have one page where they need ten. The pages do not need to be long. They need to exist and to be specific.
The cost of doing nothing this quarter
This is the part most owners react to. The math is uncomfortable but simple.
Assume that 15% of new-customer research for your category currently runs through an AI engine. By 2027, conservatively, that figure becomes 35%. If your business is invisible in 5 of 8 engines today, and you do nothing, the share of new customers you simply do not get a chance to win quietly doubles. For a dental practice adding eight new patients a month, that is a meaningful number. For a law firm adding two business clients a quarter, it is more than meaningful. For a CPA firm adding ten annual filers a season, it is a recovery line item.
The fix is mechanical and finite. It is also the kind of project that, six months from now, will have been quietly done by your direct competitors. The owners who move on it in 2026 are the ones who get to read the second half of this decade with the wind behind them.
Want the eight-engine report on your practice or firm?
We run a free 8 surface visibility report on any Seacoast NH professional services business that asks. Five business days, delivered as a PDF. No call required. No pitch attached. If the gap is fixable on your own, we will say so on the cover.
Request the report →