Google Business Profile Playbook 2026

§0 Evidence Key

Local SEO is drowning in confidently-stated nonsense. Every material claim below carries one of these labels. Never present an amber or red claim to a client as fact.

§1 Executive Summary — The Ten Things That Matter

If you read nothing else, read this. Ordered by revenue impact across a healthcare book of business.

1. Primary category is the #1 ranking lever, full stop. Whitespark's 2026 survey of 47 local SEO experts scores it 227 — ahead of proximity (225) and business title keywords (223). For a general dentist that's Dentist in ~99% of cases. Survey
2. Predefined services jumped from #81 to #22 in the ranking factors. This is the biggest riser in the study, and it's the section you asked about. Ticking Google's predefined services matters; custom services matter less; service descriptions rank #124 and are a conversion play only. Survey Tested
3. Service areas do not affect rankings. At all. Four independent tests, two firms, including one that publicly retracted the opposite conclusion. What moves geographic reach is the verified address, whether the address is shown, and map pin accuracy. Tested
4. Never hide a dental office address. "Address showing on GBP" is a top-10 ranking factor (score 176). Sterling Sky documented a local pack disappearing entirely when an address was hidden, and returning when restored. Tested
5. Stock photos are a policy violation, not merely bad practice — Google's Maps content policy explicitly bans "stock photos" and "imagery created by other parties." No documented suspensions for it alone, but they also don't help you. Official
6. Telehealth-only practices are not eligible for a GBP. "A business must make in-person contact with customers during its stated hours." No healthcare carve-out exists. Virtual offices are explicitly ineligible. Official
7. Responding to patient reviews is a live HIPAA enforcement area, and dentists are the main target. Four OCR actions ($10K–$50K); three of them dental. Even confirming someone was a patient is a disclosure. Official
8. 4.5 stars is now a hard operational floor. 31% of consumers won't use a business below it (up from 17% last year), and it's also roughly where ChatGPT's recommendation threshold sits. Survey
9. Yelp now feeds ChatGPT. As of July 2026 Yelp licenses reviews, ratings and photos to OpenAI. If you deprioritized Yelp for healthcare clients, that decision is now wrong. Official
10. GBP impressions are down roughly 54% year over year while actions held flat. AI Local Packs show fewer businesses and often strip the click-to-call button. Reporting must shift from impressions to actions, or every QBR looks like a failure. Survey Tested

§2 Services — Architecture, Limits, and the Duplicate Problem

Your original question. "Edit profile → Services" and the standalone "Edit services" panel write to the same underlying array. There is one list, rendered in category-scoped blocks.

2.1 Why you see the same service multiple times

Every service item is bound to a category ID. The editor renders one block per category on the profile — primary plus every secondary. Google auto-suggests overlapping services under each. Add five dental categories and "Teeth Whitening" can legitimately exist five times as five distinct objects.

2.2 Hard facts and limits

2.3 Two things almost nobody knows

2.4 The rebuild procedure

1. Lock categories first. Never touch services before categories are final. Every category you delete removes a duplicate service grid.
2. Kill or fix any sync tool that writes serviceItems. Otherwise it re-pushes duplicates within a day.
3. Wipe and rebuild. There is no dedupe function. Delete everything, rebuild clean.
4. One service, one home. Each real service gets exactly one entry, under its single most relevant category. Prefer Google's predefined item; go custom only when no predefined match exists.
5. Watch the justification text. Custom services render verbatim next to the practice name as "Provides: X" in the local pack. Clumsy wording is a public reputation problem. Sterling Sky documented law firms rendering as "aggravated and sexual assault."
6. Descriptions last. 300 characters, answer-first, entity-rich. Ranking value ≈ zero; conversion and AI-citation value is real.
7. Mirror to the site. GBP service name → on-site H1 → hasOfferCatalog schema. Exact string match. Entity consistency is the real GEO lever.
8. Re-audit every 30 days. Screenshot the list; Google changes it silently.

2.5 Services: do / don't

§3 Images — Specs, Stock Photos, and the Myths

3.1 Official specs

3.2 Stock photos — the actual policy

This is the finding most agencies miss, because the prohibition isn't on the GBP help pages. It's in the Maps user-contributed content policy, which GBP's photo policy explicitly incorporates:

Same policy: upload media "of a place that you captured using a camera," created "at the location you're posting to." A stock photo can additionally trip the fake engagement and misrepresentation policies.

3.3 Myths to remove from your SOPs

3.4 What photos are genuinely worth in 2026

Two real arguments, neither of which is direct ranking:

• Google replaced call buttons with images in local packs for several industries between October and November 2025. Image quality is now doing conversion work that a click-to-call button used to do. Tested
• Engagement correlation. BrightLocal's 45,000-business study: profiles with 100+ images saw 520% more calls, 2,717% more direction requests, 1,065% more website clicks vs. the average business. This is correlation, not causation — high performers invest more across the board — and the data is from 2017–18. Use it as directional colour, never as a promise. Survey

3.5 Healthcare imagery — the part with real legal exposure

3.6 Photo troubleshooting

Why photos disappear, in order of likelihood: profile not verified → still processing → flagged not-approved → incomplete core business info (missing name, address, categories, phone or hours suppresses media) → low-quality cover photo replaced by a user submission. Community-reported additions: new listings under ~14 days old are blocked from publishing; SafeSearch classifier failures (test with the Cloud Vision API SafeSearch demo); duplicate/watermarked/copyrighted detection.

Customer-uploaded photos cannot be blocked. Removal is via Report a problem only, reviewed over several business days. Reporting the same item repeatedly delays review. And note: your cover photo selection is a suggestion — Google routinely overrides it with a query-relevant or recently-uploaded image.

360 / virtual tours: deprioritize. The Street View Trusted photographer program is effectively dead (the Street View app shut down March 2023), and no credible study shows ranking or conversion lift for a small clinic. Exception: practices where facility anxiety is the conversion barrier (sedation suites, spa-style ops) — and then it's a website asset first.

§4 Service Areas — The Most Over-Sold Field in Local SEO

You flagged this as super important. It is — but in the opposite direction from what most agencies assume.

4.1 The evidence trail

Whitespark's own field tier list grades service areas F tier — "a waste of time." As of March 2026 they were still answering the question directly: "The service areas do not have any impact on rankings. All they do is show the user where you're willing to travel to."

4.2 The mechanics you do need to know

4.3 What actually determines geographic reach

1. The verified address. When the address is hidden, Google falls back to the originally-verified address as the proximity anchor — permanently. Hiding the address doesn't delete it.
2. Whether the address is displayed. Top-10 ranking factor. Sterling Sky documented a case where hiding it made the local pack disappear entirely for a main keyword, returning when restored.
3. Map pin placement. The address and the map pin are not the same thing. Proximity authority derives from the pin's geocoded coordinates. Whitespark 2026 ranks "Proper Map Pin Placement" #10.
4. Competition density and query type. Radius is query-dependent, not business-dependent. A head term in a dense suburb reaches maybe 1–3 miles; a specialty long-tail term reaches 5–15; a rural practice with few competitors can reach 20+.
5. Service-area pages on the website. This is the tactic that actually works — location pages with full NAP, community-specific content, area-specific reviews and photos, driving directions. Sterling Sky documents multi-year 3-pack rankings in distant communities from this.

4.4 Recommendation by client type

§5 Practices Without a Physical Address — Mental Health, Nutrition, Telehealth

Your side-business clients. This is the highest-risk category in your entire book, because the shortcuts everyone recommends are the exact things Google's policy was written to stop.

5.1 The three sentences that resolve almost every question

5.2 Eligibility decision tree

5.3 Video verification — how to pass it

Every video must prove three things: location (where you are), business existence (it's real), and proof of management (you run it). Google's own guidance for SABs and home-based businesses:

• Home-based businesses "can record landmarks and signs in your immediate neighborhood to verify the address."
• Businesses lacking permanent signage should "list it as a service-area business instead of storefront."
• Technical requirements: unedited, single take, no breaks, at least 30 seconds, recorded and uploaded from a mobile device. Pre-recorded footage is prohibited. Review takes up to 5 business days.

Common failure causes: stopping and restarting the recording (the single most common rejection) · the video contradicting the declared business model (declared SAB, filmed a storefront) · never showing anything only an owner could access · address in the video not matching the profile · a keyword-stuffed business name visible or declared · filming at a PO box, virtual office, or coworking space.

Pre-verification prep: have the website live at least a week beforehand with name, contact, service territory, hours and service descriptions. Build a few consistent citations first (BBB, Chamber of Commerce, professional directories). Have the license, registration, DBA, insurance and utility bills ready. After verification, wait 24–48 hours between significant edits — batching name/address/phone/category changes triggers re-verification.

5.4 Naming and categories

Category selection

Google's rule: pick categories that complete "this business IS a…", not "this business HAS a…"

Two quirks worth knowing. "Mental Health Clinic" semantically implies a premises — pairing it with a hidden address is a mismatch that raises review risk; prefer Psychotherapist or Mental Health Service for hidden-address SABs. And in states where "dietitian" is a protected title, a category/credential mismatch is a real-world compliance problem well beyond Google.

5.5 If they're genuinely not eligible — the alternative stack

Don't fake it. Build visibility on surfaces that don't require verified physical presence:

1. Vertical directories — Psychology Today, GoodTherapy, TherapyDen, Zencare, Alma, Headway, Inclusive Therapists. For therapists these are frequently higher-intent than Maps anyway.
2. State-specific service pages. Clinicians are licensed by state, so state pages are both accurate and rankable. Then go granular per modality and per presenting issue.
3. Schema markup — Organization + MedicalBusiness/Psychological, areaServed with the state list, Person schema per clinician, medicalSpecialty, credential markup. No address required.
4. Insurance panel directories — BCBS, Aetna, Optum/UHC, Cigna provider finders. Often top-3 for "therapist who takes [insurance]."
5. Bing Places and Apple Business Connect — different eligibility rules, worth testing. Apple Maps consumer usage nearly doubled (14% → 27%).
6. Unstructured citations for AI surfaces — local press, community involvement, podcast appearances, professional association pages. Three of the top five AI-visibility factors are citation-related; "mentions are the new link."

5.6 Practitioner listings for group practices

This applies equally to multi-therapist practices and multi-dentist offices. Google's rules, verbatim:

• A practitioner gets their own profile if they operate in a public-facing role and can be contacted directly at the verified location during stated hours.
• With multiple practitioners at one location: "The organization should create a Business Profile for this location, separate from that of the practitioner. The title of the Business Profile for the practitioner should include only the name of the practitioner, and shouldn't include the name of the organization."
• Support staff are not eligible.
• Where the practice name essentially is the practitioner's name ("Dr. Jane Smith Dentistry"), skip the practitioner listing — it'll be flagged as a duplicate.

5.7 Suspensions — what to know before it happens

Top causes: excessive/rapid profile edits (read as hijacking) · virtual offices, PO boxes, coworking · a compromised or restricted manager account cascading to every profile · duplicate listings · high-spam categories (including drug rehab) · online-only businesses · violations on other Google surfaces, particularly Google Ads.

Soft vs hard: soft = profile stays live but shows "Claim this business," rankings usually survive. Hard = removed entirely. Sterling Sky's finding: neither type causes lasting ranking damage if reinstated quickly — the profile returns to its prior positions. Multi-month suspensions in competitive markets are the exception.

Documentation to have ready before you need it: business registration, state license, tax certificate, utility bills at the address, lease, dated photos of signage/equipment. The appeals tool is the only channel — every other contact method redirects to it.

§6 What Actually Ranks — 2026 Factor Weights

Whitespark's 2026 Local Search Ranking Factors: 47 local SEO experts, 187 factors, scored 0–5. New this year: factors are also scored for AI Search visibility. Scores are relative weights, not percentages.

6.1 Top 20, Local Pack / Maps

Top 5, Localized Organic

1) Dedicated page per service (210) · 2) geographic keyword relevance (190) · 3) quality/authority of inbound links (187) · 4) keywords in GBP landing page title (179) · 5) links from industry-relevant domains (175).

Note #1: a dedicated page per service outranks everything else on the organic side. This is the strongest argument for building out individual treatment pages and mirroring them to GBP services.

6.2 Things that do not rank (stop selling them)

6.3 Operational gotchas most agencies miss

§7 Reviews — Consumer Reality, and the HIPAA Trap

7.1 BrightLocal Local Consumer Review Survey 2026 (n=1,002 US adults)

Also: 78% of reviewers were asked, and 65% of those complied. Asking works. 19% expect a same-day owner response (up from 6%); 81% expect one within a week.

7.2 HIPAA — the highest-liability item in this entire document

Merely confirming or denying that someone is or was a patient is a disclosure. "Thanks for being a great patient, Susan!" on a five-star review is a violation. This applies to positive reviews too, which is where most practices get blindsided.

Three of the five are dental practices. This is a live, repeatedly-enforced risk in your exact vertical.

Note the ADA petitioned the FTC in January 2023 for a healthcare carve-out allowing disclosure when rebutting deceptive reviews. It was not granted. And: "Just because a patient identifies themself in a review, they have NOT waived their right to privacy."

7.3 Review gating is now double jeopardy

Google side: gating is classified under Deceptive content. Reported consequence is removal of your entire review profile, not just the gated reviews. Google's policy also prohibits offering incentives, requiring or pressuring users to review while on the premises, requesting specific content, and reviews from employees or contractors.

Federal side: the FTC Consumer Reviews and Testimonials Rule (16 CFR Part 465) took effect October 21, 2024 and carries civil penalty authority for first-time violations, which the FTC previously lacked. It bans fake and AI-generated reviews (§465.2), incentives conditioned on sentiment (§465.4), undisclosed insider reviews (§465.5), company-controlled "independent" review sites (§465.6), and review suppression via threats or intimidation (§465.7).

§8 AI, GEO & AEO — Where Local Discovery Is Actually Going

8.1 The visibility gap is brutal

SOCi's 2026 Local Visibility Index analyzed ~350,000 locations across 2,751 multi-location brands:

Search Engine Land's summary: "AI visibility is three to 30 times harder to achieve than ranking well in traditional local search." Local Falcon's February 2026 restaurant study (189,905 ChatGPT results) found 83% of restaurants completely invisible on ChatGPT vs. 14% on Google — 10× worse for US businesses.

8.2 Each engine has a different data supply chain — optimize accordingly

8.3 What Google AI Mode draws on

Per Whitespark's May 2026 guide: GBP (ratings, hours, contact info) · review platforms including Yelp · unstructured citations (blogs, news, community pages, social) · website content and FAQs · industry publications and sponsorship pages.

Mechanics that matter: query fan-out — the system issues multiple related sub-queries rather than one. Results are personalized, so every user sees different results for identical queries. Being in the top 10 organic gives only about a 25% chance of appearing in AI Overviews. AI Overviews appeared in roughly 68% of local searches in Q1 2025, and 67% of consumers don't fact-check AI sources.

8.4 The Q&A section is gone — and what replaced it changes the play

8.5 Local AEO/GEO action list

1. Get every client above 4.5 stars. This is now a gating threshold for AI recommendation, not just a conversion nicety — ChatGPT's recommended locations averaged 4.3, above the 3.9 that suffices in traditional search.
2. Rebuild Yelp profiles for healthcare clients. Complete, photographed, actively reviewed.
3. Write website FAQs in conversational language — the exact phrasing patients use when talking to an assistant, not keyword-optimized headers.
4. Build unstructured citations — local press, community sponsorship, professional associations, podcast appearances. Three of the top five AI-visibility factors are citation-related; "mentions are the new link."
5. Fix NAP everywhere. The 68% accuracy figure on ChatGPT and Perplexity is a citation-hygiene problem, and it's fixable.
6. Enroll in Reserve with Google — explicitly named as an AI-visibility optimization. Dental integrations exist via NexHealth, Planet DDS, Dental Intelligence, Zocdoc.
7. Answer the aggregated Ask Maps questions thoroughly. That's now AI input, not page furniture.
8. Change what you report. Impressions are down ~54% YoY while actions held roughly flat. AI Local Packs show fewer businesses (32% fewer unique businesses surface) and often drop the click-to-call button entirely. Report actions, calls and booked appointments — not impressions — or every QBR reads as a failure that isn't one.

§9 Regulatory Landmines Specific to Dental & Medical

9.1 The "specialist" advertising problem — the most under-appreciated exposure in dental GBP work

The popular tactic is to add the Orthodontist category to a general dentist's profile to capture Invisalign searches. Sterling Sky openly recommends it, noting "you don't technically have to be an orthodontist to do Invisalign" — while adding that it should be approved by the dentist first. That caveat is doing a lot of work.

9.2 Business name keyword stuffing — the uncomfortable truth

Sterling Sky tracked 50 keyword-stuffed listings in the legal vertical, reporting each and escalating repeat offenders. Outcomes: 60% warning only · 20% soft suspension · 20% hard suspension. 100% of subjects re-added the keywords after being caught; one cycled eight times without a permanent ban. Their conclusion: "Google lacks automated systems to catch keyword stuffing; enforcement is almost entirely manual and triggered by competitor reports."

Honest framing: it works, in the sense that it lifts rankings (business title keywords are the #3 factor) and is rarely auto-caught. It is also a guidelines violation, trivially competitor-reportable, and the downside tail is a hard suspension that kills the profile. In dental specifically it compounds into a state board advertising problem if the stuffed keyword implies specialty status. Our position should be: don't do it, and file redressal complaints against competitors who do.

9.3 Fighting competitor spam

The Business Redressal Complaint Form accepts: title/name spam, address spam (virtual offices, UPS Store addresses), phone number spam, website spam, and listings for non-existent businesses. It rejects anything not related to fraudulent activity on the name, phone number or URL — don't send it ranking complaints or info corrections.

Evidence that works: government sources (Secretary of State, state dental and medical licensing boards — very strong in your vertical), Google Street View screenshots, links to the legitimate business at that address. Avoid Facebook and Yelp as evidence. Host images on Google Drive; don't embed them in the spreadsheet. Use a spreadsheet upload for 10+ listings. Expect ~2 weeks; you get a case ID but cannot track progress.

9.4 Local Services Ads for healthcare — and a terminology change you need to fix in your decks

Healthcare is fully supported in the US: Dentist, Orthodontist, Dermatologist, Plastic surgeon, Orthopedic surgeon, Ophthalmologist, Optometrist, Podiatrist, Primary care physician, Chiropractor, Acupuncturist, Audiologist, Dietitian, Occupational therapist, Physical therapist, Veterinarian.

Healthcare-specific LSA limits: booking leads aren't available for healthcare verticals (call and message leads only), and pre-badge ads aren't available — healthcare cannot run LSAs before completing full verification. No shortcut to launch. Also coming: LSAs are migrating into the main Google Ads platform through 2026, and D-U-N-S numbers are being required for some advertisers as of July 2026.

Paid search benchmarks for context (WordStream 2026, 13,474 US search campaigns, medians): Dentists — 5.66% CTR, $8.00 CPC, 10.67% CVR, $72.97 CPL. Physicians & surgeons — 6.61% CTR, $4.76 CPC, 12.43% CVR, $40.04 CPL (down 29.5% YoY). No credible published LSA-specific cost-per-lead data for dental exists — anyone quoting one is guessing.

§10 Platform Changes 2024–2026 — What Broke, What's New

Best single source to monitor continuously: Sterling Sky's Google Local Changes log.

§11 Build SOP & Recurring Audit

11.1 New client GBP build — sequence matters

1. Eligibility check. In-person contact during stated hours? Real staffed address with signage? If telehealth-only or virtual office — stop, go to the §5.5 alternative stack.
2. Categories. Primary = what the business is. Fill secondaries with what they genuinely provide. Screen every specialty-implying category against state board rules.
3. Core NAP + hours. Real practice number as primary. Accurate hours, including holidays — "open at time of search" is the #5 ranking factor.
4. Map pin verification. Drop it on the actual entrance. Address ≠ pin.
5. Services. Predefined first, one home per service, custom only for real gaps. Then 300-char descriptions.
6. Products. 8–12 merchandised treatment cards with real photos, prices where the practice will honor them, and Learn More links to matching service pages with UTMs.
7. Photos. Original only. 3+ each of exterior, interior, team, work. Never stock. Wait 14 days post-verification before bulk upload.
8. Attributes. Accessibility, languages, insurance, appointment/booking, identity attributes where accurate.
9. Booking integration — and remove any unauthorized third-party booking link already there.
10. UTM-tag every link on the profile.
11. Mirror to the website — service page per service, matching H1s, hasOfferCatalog schema, NAP in HTML matching GBP exactly.
12. Review system. Compliant ask (everyone, no gating, no incentives) + pre-approved HIPAA-safe response templates.
13. Baseline geogrid. 9×9 or 11×11 at 1.5–2 mi for head terms; second scan at 5–7 mi for specialty terms. Track share of grid in top 3, not average rank.

11.2 Monthly audit checklist

11.3 Honest gaps — do not fabricate these for clients

• Med spa / derm / plastic surgery primary category testing — no published study exists anywhere. Treat category choice there as a testable hypothesis (geogrid A/B over 4–6 weeks), not a known best practice.
• Multi-location dental cannibalization — no study; agency blogs only. "Unique content per location page" is consensus, not proven.
• Current dental review count / rating benchmarks — the only segmented data is BrightLocal's 2017–18 study (dentists: median 39 calls, 28 direction requests, 46 site clicks, 12 photos per month). It predates GBP's Insights rebuild and the AI shift. Benchmark each practice against its own trailing 12 months and its top 3 geogrid competitors instead.
• LSA cost-per-lead for dental — no credible published figure.
• Numeric ranking radius by vertical — no study exists. Anyone quoting one is guessing.
• Telehealth / "online care" attributes and "accepts new patients" in 2026 — unverified. Check a live dashboard before promising a client the field exists.