Why AI search decides differently than Google

A Google search for “dentist in Brooklyn” returns ten blue links, a map, and four ads. The patient scans, clicks, compares, and decides. An AI search for the same thing returns a paragraph: “If you are in Brooklyn, three well-reviewed practices are [Name A], [Name B], and [Name C], each offering…” The patient often does not click anything. They screenshot, call, or book.

This is already mainstream behavior. According to KFF's June 2026 tracking poll of 2,480 US adults, 29% now use AI tools or chatbots for health information or advice at least monthly, nearly double the 17% KFF measured in June 2024. And reviews still sit at the center of the choice: in a Dentaly survey of 1,099 Americans (October 2023), 83% said they use online reviews to evaluate a dentist, and 52% would only consider a dentist with at least 10 online reviews.

The shift matters because the pick is narrower. Google shows ten options. In our prompt testing, ChatGPT and Perplexity usually pick three. Copilot sometimes picks one. That compression pushes all of the marketing weight onto a single question: what does the engine already believe about your practice, right now, before the patient asked?

AI engines build that belief by reading the public web and cross-referencing what multiple trusted sources say about you. They do not run live auctions like Google Ads. They do not rank by keyword match like classic SEO. They pattern-match on consistency, authority, and specificity, and they do it before the patient's prompt is ever typed.

Google is explicit that there is no secret AI track. Its Search Central guidance on AI features says: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” The same page explains that AI Mode uses a “query fan-out” technique, running related searches across subtopics and data sources, and it tells site owners to keep their Business Profile information up to date. For a dentist, that means the fundamentals below are the AI strategy.

The seven signals AI engines use to pick a dentist

1 Review volume and consistency across platforms

Google states it plainly in its Business Profile help page, Tips to improve your local ranking on Google: “More reviews and positive ratings can help your business's local ranking.” AI answers that draw on Google's local data inherit that weighting.

WebFyndr internal benchmark across 40+ dental clients (our own data, not a published study): practices named first in ChatGPT recommendation prompts averaged 147 Google reviews, 18 Healthgrades reviews, and 23 Yelp reviews, with a 4.6+ star average on all three. Practices below 50 Google reviews were named in fewer than 8% of the prompts we monitor.

Volume alone is not enough. In our experience, steady review velocity (at least 3 new reviews per month, not one burst followed by silence) outperforms a one-off campaign. Platform spread matters too, because patients cross-check: BrightLocal's Local Consumer Review Survey 2025 (1,026 US adults) found 74% of consumers use two or more review sites. For dentists specifically, Dentaly's survey found Google is the most trusted review platform (85%), followed by Yelp (37%) and Healthgrades (25%). A practice with 200 Google reviews and zero Healthgrades presence looks thin to a patient and to an engine reading across sources.

2 Schema.org markup for LocalBusiness + Dentist

Use the specific type. On schema.org, Dentist sits under both LocalBusiness (via MedicalBusiness) and MedicalOrganization, so a single Dentist block inherits every LocalBusiness property. Google's LocalBusiness structured data documentation gives the same advice: use the most specific LocalBusiness sub-type possible. Its only required properties are name and address; recommended ones include geo, openingHoursSpecification, telephone, and url.

Fill that block with full NAP (name, address, phone), opening hours, geo-coordinates, sameAs links to your Google, Healthgrades, and social profiles, plus services, specialties, and languages spoken. Keep it honest: Google's AI features guidance says structured data should match the visible text on the page. And skip self-hosted star markup: Google announced in September 2019 that self-serving reviews are not allowed for LocalBusiness and Organization, so stars from your own testimonials widget will not show as review rich results.

In WebFyndr audits, fewer than 25% of the US dental practice sites we review have valid Dentist schema. In our experience, adding it in a single afternoon moves some practices from “unknown to the AI” to “plausible candidate” within two to three weeks.

3 Citation consistency across directories

Name, address, and phone must match exactly across Google Business Profile, Healthgrades, Yelp, Zocdoc, WebMD, 1-800-Dentist, your state dental association directory, and the ADA Find-a-Dentist tool. One suite number difference or a hyphenated phone format mismatch flags inconsistency and lowers the engine's confidence.

Audit in a spreadsheet quarterly. The single fix that moves the needle for practices that have moved locations or changed phone numbers in the last five years is a directory cleanup pass.

4 Depth of content that demonstrates clinical expertise

Thin sites with “Services: Cleanings, Fillings, Crowns” get no traction. Sites with treatment-specific pages, an implant page that explains zirconia vs titanium, recovery timelines, cost ranges, and what makes your case selection conservative, signal expertise to the engine. Each specific, true claim (for example, a dentist stating how many implants they place per year and their own documented survival rate) becomes a potential pull-quote.

There is research behind this. In “GEO: Generative Engine Optimization” (Aggarwal et al., Princeton and IIT Delhi, KDD 2024), the authors report that optimization methods “can boost visibility by up to 40% in generative engine responses.” The strongest methods were adding quotations, adding statistics, and citing sources, while keyword stuffing performed below the unoptimized baseline.

Aim for 800 to 1,500 words per major treatment page, with structured sub-headers that mirror the questions patients actually ask.

5 Third-party mentions on trusted sources

A single interview in Dentistry Today, a guest post on Dentaltown, a quote in a local newspaper about a cosmetic procedure, or a guest appearance on a dental podcast, each of these shows up when the engine searches for your practice name. Mentions from sources the engine already trusts carry disproportionate weight.

We aim for two new third-party mentions per quarter for each dental client. Over 12 months that is 8 external validations, which in our client work is often the difference between a top-3 and top-10 AI recommendation position.

6 Google Business Profile completeness and activity

A complete GBP has 15+ photos of the office and team, every service populated, weekly posts (updates, offers, events), messaging enabled, and attributes filled in (wheelchair accessible, online booking, accepts insurance). Google's own local ranking tips say: “Businesses with complete and accurate info are more likely to show up in local search results.”

The profile now feeds AI answers directly. Google retired the old Business Profile Q&A feature (its API was shut down on November 3, 2025) and replaced it with Ask Maps, a Gemini-powered tool in Google Maps. According to BrightLocal's April 2026 breakdown, Ask Maps builds answers from your Business Profile (services, attributes, hours, photos, posts), from reviews, and from your own website's service pages and FAQs. A practice with a half-filled GBP gives the AI less to say, even if its website is strong.

7 Site freshness and content velocity

A dental site that has not been updated in 18 months signals neglect to the engine. The fix is modest: one new patient-education article per month, updated treatment pages once per quarter, a quarterly “what's new at the practice” post. Nothing heroic, just demonstrated ongoing activity.

The 90-day GEO plan for a dental practice

Below is the exact cadence we run. It is not a maximum program, it is the minimum sequence that consistently produces measurable AI visibility lift in the fastest practices we work with.

Weeks 1–4

Foundation

  • Audit all 12 major directories and fix NAP inconsistencies
  • Deploy LocalBusiness + Dentist schema
  • Complete Google Business Profile (photos, services, attributes, description)
  • Baseline AI visibility audit across ChatGPT, Perplexity, Gemini, Copilot for 12 target prompts
  • Set up review-ask automation (post-visit text with one-tap Google link)
Weeks 5–8

Authority

  • Publish 4 treatment-specific pages (800–1,500 words each)
  • Place one guest post on Dentaltown or a state dental-society blog
  • Secure one trade-press mention (Dentistry Today, Inside Dentistry)
  • Push 20+ patient review asks per week
  • Weekly GBP posts; respond to every review within 48 hours
Weeks 9–12

Monitoring

  • Re-run the baseline audit, track prompt coverage delta
  • Identify 3 prompts where you are named and 3 where you are not, find the gap
  • Publish 2 more content pieces to close the gap
  • Set up monthly tracking cadence (15 min per month to log citations)
  • Document wins in a case study for your own PR use

Case: Phoenix pediatric practice, 0 to top-recommended in 84 days

Illustrative case based on WebFyndr client work, details generalized to protect practice identity

Starting point (Day 1): 62 Google reviews at 4.4 stars. No Dentist schema. Healthgrades profile abandoned 2 years prior with wrong phone number. Website last updated 2023. ChatGPT recommendation prompt “pediatric dentist in [area of Phoenix]” named three competitors and did not mention this practice. Perplexity: same. Gemini: same.

What we did: Directory cleanup (9 profiles corrected or claimed). LocalBusiness + Dentist + MedicalBusiness schema deployed. GBP rebuilt with 47 photos, full service list, and weekly posts. Six treatment pages written (sedation, fluoride alternatives, thumb-habit appliances, emergency visits, first-visit guide, insurance-acceptance FAQ). One interview placed in a regional parenting publication. Post-visit text sequence pushed 212 review asks in 12 weeks; 94 new Google reviews landed.

Result at Day 84: 156 Google reviews at 4.7 stars. Named in 9 of 12 monitored AI prompts (Day 1: 0). Patient intake form asks added a “how did you find us” field, 17% of new patients in weeks 10–12 said “ChatGPT” or “an AI told me.” Practice revenue up 23% year-over-year in Q2, with the owner attributing the majority to digital inbound rather than referrals.

What fails more often than it succeeds

Buying reviews. Fake reviews tend to be filtered quickly; the pattern (IP clusters, no profile history, templated language) is obvious. It is also illegal: the Federal Trade Commission's final rule (announced August 14, 2024) bans buying fake reviews and bars “compensation or other incentives conditioned on the writing of consumer reviews” expressing a particular sentiment. Google's own review tips say incentives such as free or discounted services in exchange for reviews are “strictly prohibited.”

Keyword-stuffing the homepage with city names. AI engines read across the site and across the web, and the Princeton GEO study found keyword stuffing did worse than doing nothing. A single well-structured Dentist schema block beats “Best Family Dentist in Houston, Texas, USA” repeated 14 times in the footer.

Ignoring the front desk. Patients who had a bad front-desk experience write bad reviews no matter how good the clinical care was. Front-desk training is a GEO investment dressed in operations clothing.

Replying to reviews like a restaurant. HIPAA applies to your responses. The ADA News (March 2025) reports that in an ADA Health Policy Institute survey, 88% of dentists had received online reviews, 52% had responded, and 39% said they were unable to respond to a review because of HIPAA. The ADA's first rule: “Don't acknowledge that a reviewer was a patient, whether the review is positive or negative.” Reply in general terms and invite the reviewer to call the office.

Posting generic stock-photo “tips” on social. In our testing, an Instagram grid rarely shows up as a source in AI answers about dentists. Third-party articles, review sites, and your own treatment pages do. Social is a channel for patient trust, not a primary AI signal.

How to measure whether GEO is working

Three metrics, reviewed monthly:

Prompt coverage. Pick 12 prompts that a patient in your area might actually type. Run each through ChatGPT, Perplexity, Gemini, and Copilot. Log how many name your practice, and in what position. Baseline at Day 1, re-measure at Day 30, 60, 90, then monthly.

Citation source list. When an AI engine cites your practice, which source did it pull from (your site, Healthgrades, a news article, Yelp)? Over 90 days you want the list to lengthen and diversify. A practice cited only from its own site is fragile.

New-patient attribution. Add “Where did you hear about us?” to your intake form. Add “ChatGPT / AI search” and “Perplexity / other AI tool” as explicit options. Patients who answer honestly give you the ground truth no dashboard can provide.

What this means for a typical US dental practice

If you run a 2–5-chair practice with $900K to $3M in annual production, the math on GEO is straightforward. As an illustrative WebFyndr estimate (not an industry figure), a single new patient from a cosmetic or implant consult can be worth $3,000 to $15,000 in lifetime value. You need the plan to deliver two to three incremental new patients per month to pay for itself at any reasonable agency fee. In WebFyndr client work, dental practices typically see 8 to 20 incremental new-patient inquiries per month attributed to AI search by month 4, measured through intake-form attribution.

The catch is that the window is narrowing. As competitors build their schema, populate their directories, and publish depth-of-expertise content, the work to catch up compounds. The practice that deploys the foundation layer this quarter wins a head start over the practice that waits until next year. A year from now, the cheap wins will be gone and only the incremental work will be left.

Frequently Asked Questions

Why does ChatGPT recommend some dentists and not others?

AI answers reflect what the web says about you: brand mentions on trusted sources, review volume, Schema.org coverage, and citation consistency across directories. In our prompt testing, a practice with 180 Google reviews, full Dentist schema, and three trade-press interviews gets named in most recommendation prompts; a practice with 40 reviews and a bare site does not. It is not random and not a simple ranking winner. It is the sum of seven signals working together.

How many reviews does a dental practice need for AI visibility?

There is no official number. WebFyndr's internal working benchmark is at least 80 Google reviews with a 4.6+ star average, plus 3 or more new reviews per month. In our client data, practices below that threshold are rarely named first, and above 150 reviews recommendations increase sharply, especially when paired with Healthgrades or Yelp presence that matches your Google profile.

Are US dentists allowed to actively ask for patient reviews?

Yes, as long as the ask is neutral and the patient is free to write whatever they want. The ADA's practice guidance, Protect your practice's reputation, says dentists can encourage patients to leave truthful reviews, but warns that offering a gift or prize chance in exchange for a review can trigger disclosure rules or be prohibited, and that some jurisdictions restrict soliciting patient reviews. The FTC's 2024 rule bans incentives conditioned on a particular sentiment, and Google prohibits incentives for reviews outright. Keep the ask simple: a post-visit text or email with a direct review link, no rewards, no scripted wording.

How much time does GEO take for a dental practice?

A typical 90-day engagement takes the practice itself about 2 to 4 hours per month: interviews about case outcomes, content approval, and small website edits. The technical work (schema, directories, monitoring, content, PR) is done by the agency. After 90 days it drops to 1 to 2 hours per month for maintenance.

Can a dental practice combine GEO with existing marketing spend?

Yes. Most practices already have a budget for Google Ads, Meta ads, and perhaps a local agency. GEO does not replace any of that. In WebFyndr client work, 20 to 30 percent of the existing SEO budget typically shifts toward GEO-specific activity without increasing total marketing spend.

Related Reading

Sources

  1. KFF Tracking Poll on Health Information and Trust: Use of Social Media and AI For Health Information and Advice, KFF, June 2026.
  2. Poll: Most Who Use Artificial Intelligence Doubt AI Chatbots Provide Accurate Health Information, KFF, August 2024.
  3. 8 Out of 10 Americans Choose Their Dentist Based on Online Reviews, Dentaly.org, survey fielded October 2023.
  4. AI Features and Your Website, Google Search Central, updated December 2025.
  5. Tips to improve your local ranking on Google, Google Business Profile Help, accessed October 2026.
  6. Local Consumer Review Survey 2025, BrightLocal, January 2025.
  7. Dentist, Schema.org, accessed October 2026.
  8. Local business (LocalBusiness) structured data, Google Search Central, accessed October 2026.
  9. Making Review Rich Results more helpful, Google Search Central Blog, September 2019.
  10. GEO: Generative Engine Optimization, Aggarwal et al., arXiv 2311.09735 (KDD 2024), June 2024.
  11. Goodbye Google Q&A: It's Time to Ask Maps, BrightLocal, April 2026.
  12. Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials, Federal Trade Commission, August 2024.
  13. Tips to get more reviews, Google Business Profile Help, accessed October 2026.
  14. 5 best practices for responding to online reviews, ADA News (American Dental Association), March 2025.
  15. Protect your practice's reputation, American Dental Association, accessed October 2026.