How Dental Patients Find a New Practice in the Search-and-Answer Era

The patient runs a sequence, not a search

It is tempting to picture new-patient acquisition as a single query with a winner. The evidence on how people choose a provider points to something more procedural: a filter that removes candidates in stages until one name is left standing. Each stage answers a different question, and a practice can be eliminated at any of them without ever being compared on the things it is proudest of.

The useful frame is that a dental patient is buying a credence good, something whose quality they cannot verify before purchase and often not even after. You can taste a meal and time a plumber, but you cannot assess a dentist's clinical judgment by looking at the website. Economists have long noted that credence-good buyers substitute observable proxies for the quality they cannot inspect, and the search-and-answer era has simply made those proxies faster to gather and harder to fake casually. Reviews, in-network status, distance, and visible credentials are exactly those proxies. The sequence is the buyer assembling them in roughly ascending order of effort.

Two caveats belong here at the start, not buried at the end. First, the order is discernible but not universal: an emergency toothache collapses the sequence toward "nearest, open now, takes my plan," while a cosmetic consult stretches it toward credentials and before-and-after proof. Second, and more important, no single authoritative study of the dental new-patient journey specifically was located for this piece. The sequence below is built from established provider-choice and local-search evidence applied to dentistry, and we flag every place that inference carries weight the primary data does not.

Location: the map pack decides who gets considered

For a local, recurring, in-person service, physical proximity is a hard filter, and the surface that expresses it is Google's local pack, the boxed set of three businesses on a map that sits above the ordinary links. Its share of local-intent attention is disproportionate to its three slots.

Aggregated local-search behavior research reports that searchers click the local three-pack around 44 percent of the time, against roughly 29 percent for organic links and 19 percent for paid, and that the top pack position draws a materially larger share than the second or third. Businesses in the pack are reported to receive substantially more traffic and more direct actions, calls, direction requests, and site clicks, than comparable businesses left out of it. The direction of this finding is well established across the local-SEO literature; the exact percentages are secondary-sourced rather than drawn from a single disclosed methodology, and we treat the magnitude as emerging rather than settled.

The operational point survives the uncertainty. Being absent from the pack for "dentist near me" and its procedure-and-plan variants is not the same as ranking a little lower. It is being outside the consideration set the patient assembles, before reviews or credentials are ever weighed. Distance is the gate, and the pack is where the gate is drawn.

Reviews: the pivot check where the choice is usually made

Once a patient has two or three nearby candidates, reviews do the discriminating. This is the step with the strongest independent evidence, and it is worth being precise about what that evidence shows and where it comes from.

Insurance: the in-network directory is its own search surface

For most routine dental care, a plan is involved, and "do you take my insurance" is a near-disqualifying question asked early. The surface where it is answered is not the practice's own site; it is the insurer's in-network provider directory, the filtered list a patient generates inside their Delta Dental, Cigna, MetLife, or Aetna account, alongside cross-listings on healthcare directories an engine can pull from.

We could locate no disclosed primary study quantifying how often a wrong or missing in-network listing costs a dental practice a booking, so we make no numeric claim here. What is defensible is structural: a directory that lists the wrong phone number, an outdated address, or a lapsed participation status silently routes a ready-to-book patient elsewhere, and the practice never sees the lost call. Identity consistency, the same name, address, phone, provider roster, and accepted plans across every directory, is what lets both insurers and engines resolve the practice to one trustworthy record rather than hedging between conflicting ones.

Credentials: verifying the quality the patient cannot inspect

The last and highest-effort check is legitimacy: is this provider real, licensed, and appropriate for the procedure in question. This is the credence-good problem in its sharpest form, and there is direct evidence that patients struggle with it.

A study of provider selection in aesthetic medicine found that patients routinely cannot verify the credentials that actually govern safety, and that they frequently cannot distinguish a properly qualified provider from one who is not. Dentistry is less severe than unsupervised cosmetic medicine, but the same asymmetry holds: a prospective patient cannot assess clinical competence directly, so they lean on the visible proxies, board or specialty language, provider bios, professional affiliations, and the structured signals a machine can read. Where those proxies are absent or inconsistent, the patient and the engine both fall back to whatever they can resolve with confidence, which is often a competitor with cleaner signals rather than better care.

This is also where the honesty of the surrounding regulation matters. Because the buyer cannot verify quality, society substitutes rules: professional-conduct standards, and, for reviews specifically, the FTC's trade regulation rule on consumer reviews and testimonials, effective October 21, 2024, which makes fake and deceptive reviews a specified unfair-or-deceptive act. In the exact verticals where the buyer has no independent way to check, buying, boosting, or fabricating the trust signal is both against the rules and against the patient's interest. Credential-accurate signals are the only kind that function.

The answer layer now sits inside the sequence

A growing share of patients no longer assemble these checks by hand across four surfaces. They ask an assistant to do it, and read back a synthesized answer that already blends location, reviews, and apparent fit into a short list of names.

The adoption shift is real and fast. One widely cited annual consumer survey reported that 45 percent of consumers had used an AI tool such as ChatGPT, Gemini, or Perplexity to find a local-business recommendation in the trailing year, against 6 percent a year earlier; that single-source jump is large enough that we label it emerging and worth independent verification. The trust picture is more sober: Pew Research Center's 2026 survey found that while 49 percent of US adults now use chatbots at all, up from 23 percent in 2023, only 29 percent of chatbot users trust the information they get "a lot" or "some." Patients are using the answer layer to shortlist, then verifying it on the older surfaces, which means a practice has to be present in both.

And the two systems do not agree. Vendor research indicates that ranking in Google's local pack is a poor predictor of being named in an AI local recommendation, with the AI surface reported as markedly harder to earn than the map pack. That divergence is single-vendor sourced and should be read as directional, not precise, but it fits the broader pattern this publication tracks: being indexed and ranked is one job, and being cited inside an answer is a different one.

Each check is a surface you can win or lose

The reason to map the sequence is that it is actionable. Each stage corresponds to a distinct surface with distinct, engineerable signals, and a practice that is losing new patients is almost always losing them at an identifiable one of them, not at all four.

Read as a system, the four checks compose one measurable question: across the surfaces a patient now uses to filter, where does this practice actually stand today. That is what a visibility read establishes before any work is scoped, and it is the necessary starting point, because the correction for a missing map-pack presence is not the correction for a thin review corpus or an inconsistent insurance listing.

• Location, the map pack: a complete Google Business Profile with the correct primary category, accurate service area, and the procedure-and-plan variants patients actually search.
• Reviews, the aggregate and its text: a compliant system that earns reviews from real patients and surfaces the procedure language engines read as proof of scope, answered without breaching patient privacy.
• Insurance, the in-network directories: one consistent identity, accepted-plan accuracy, and correct contact details across insurer finders and healthcare directories.
• Credentials, the trust proxies: provider-level and service-level structure, accurate affiliations, and credential-accurate schema so both patients and engines can resolve who the practice is and what it does.
• The answer layer, across all four: presence and consistency clean enough that an AI assistant can name the practice with confidence, not just index it.

What the evidence does not say

The tier assigned to each claim is part of the method. The strongest evidence here, provider choice responds to narrative review content, and ratings move revenue for independents, is established and peer-reviewed, but neither study is dental-specific. The local-pack click behavior is established in direction and only emerging in exact magnitude. The AI-adoption jump and the pack-versus-AI divergence are single-source and emerging.

There is also a specific figure we deliberately do not cite. Marketing content across the industry repeats percentages such as "70 percent of patients use the internet to find a new dental practice" and "77 percent cite reviews as the first step," attributed to the American Dental Association. We were unable to trace these to a disclosed primary ADA study; a direct check of the ADA's own materials found patient-survey methodology guidance, not published national findings of that kind. Repeating those numbers as fact would be a fabrication risk, so this piece does not use them. The sequence stands on the evidence that exists, and the places it does not is exactly where a practice's own measured surface, not a borrowed statistic, is the thing worth reading.