AEOGeoAI Research Report · July 2026
Miami Health AI Search Visibility Study 2026Analysis of 298 Health Practices Across ChatGPT, Claude & Gemini
An Independent Analysis of AI Citation Patterns Across 298 Miami-Dade Health Practices.
An Independent Analysis of AI Citation Patterns Across 298 Miami-Dade Health Practices.
ChatGPT, Claude, and Gemini almost never failed to answer local health questions in Miami. They answered by naming competitors while omitting the tested health practice.
The competitors that do get named are overwhelmingly large institutions — not comparable independent health practices. See Section 3.
This report is part of the larger Miami AI Search Visibility Study (515 businesses across health, legal, and real estate). It documents observed AI citation patterns for the health sector specifically, tested during July 2026.
This study measures AI citation presence — defined as whether a health practice is:
This study measures AI citation presence rather than Google rankings, website quality, reputation, or business tenure. For businesses evaluating their own standing, the free AI Search Visibility Checker applies this same methodology on demand.
These findings should not be interpreted as measurements of:
Instead, they measure only AI citation presence — whether ChatGPT, Claude, or Gemini named a specific business by name — during the study period in July 2026.
Each health practice was queried individually across three AI systems using a standardised prompt. We tested whether the AI system included the named business in its recommendation-style answer. See also the full AEOGeoAI methodology page and the Miami AI Search Visibility Study for the combined cross-sector methodology.
This study measures observed output behaviour rather than internal model architecture. No conclusions are drawn regarding the proprietary ranking or retrieval mechanisms of ChatGPT, Claude, or Gemini beyond the responses produced during testing.
"best [category] in [city] FL"
Business selected
↓
Prompt generated ("best [category] in [city] FL")
↓
Query sent independently to ChatGPT, Claude, and Gemini
↓
Response collected from each model
↓
Automated secondary classifier scores each response
↓
Citation score assigned (0 / 1–34 / 35+) per model
↓
Highest score across the three models used as the business's headline result
Each business's category was used directly, or reduced to its primary field where a compound category was listed. Example: a Coral Gables dermatology practice received the query "best dermatology in Coral Gables FL".
A binary name-match alone treats "AI discussed the right category and named three competitors, but not this business" identically to "AI has never heard of this business or its category" — both would score zero. This study distinguishes the two. A raw score of 35 or higher indicates the business's name was actually matched in the response — a genuine citation. Scores of 1–34 indicate category-adjacency. A score of 0 means no relevant signal was found at all.
The 35-point threshold marks the point at which the classifier identified the business's own name as explicitly present in the response text, rather than only its category or general attributes being discussed. AEOGeoAI has not published a formal statistical derivation of this specific cutoff; it functions as a categorical boundary within our scoring system rather than a value derived from this dataset.
| Score | Meaning |
|---|---|
| 0 | No relevant signal — category not discussed |
| 1–34 | Category-adjacent — topic/competitors discussed, business not named |
| 35–59 | Genuine citation — business named, limited prominence |
| 60–79 | Genuine citation — consistently included |
| 80–100 | Genuine citation — prominently featured |
Health Practices were identified from commercially sourced contact lists, used only to identify candidate businesses to query — not to influence results. Commercial contact lists were used solely to identify businesses for inclusion; they were not used as sources for scoring, ranking, or classification. No business paid for inclusion in this study, and no business was excluded based on its citation outcome.
| Parameter | Value |
|---|---|
| Report ID | AEOGEOAI-MIA-HEALTH-2026-001 |
| Dataset ID | MIA-HEALTH-298-v1 |
| Sample size | 298 health practices |
| Top categories tested | Dentistry (19), Medical spa/aesthetics (14), Physical therapy (24 combined), Dermatology (7), Primary care (12 combined) |
| Top cities | Miami (120), Fort Lauderdale (24), Coral Gables (19), Hollywood (16), Miami Beach (13) |
| Test period | July 2026 |
| Rate control | 4-second delay per query |
| Tooling | AEOGeoAI visibility scoring system with automated secondary classifier for category-adjacency and competitor extraction |
Individual business names are not published in this report. The underlying dataset is structured as follows:
| Column | Description |
|---|---|
| Business | Business name (withheld in this published report) |
| City | City used in the query |
| Category | Category used in the query |
| Claude score | 0–100 |
| Gemini score | 0–100 |
| ChatGPT score | 0–100 |
| Classification | True zero / Category-adjacent / Genuine citation |
| Competitors named | De-duplicated list of businesses named in place of the tested business |
The query format, scoring thresholds, and category-adjacency detection logic are documented in this section. Independent researchers with access to ChatGPT, Claude, and Gemini can reproduce the general approach using the identical query format and thresholds described above. Full row-level data is available on request — see the FAQ below.
Limitations: AI outputs are probabilistic and may vary across time and model updates. Results represent a snapshot of observed behaviour in July 2026 and should not be interpreted as permanent or universal. This study only tested ChatGPT, Claude, and Gemini — it does not include Google AI Mode or Google AI Overviews specifically.
Scope of these findings: this study describes the observed sample of 298 Miami-Dade health practices and is not intended to estimate AI citation prevalence across all health practices in Florida.
Scoring definitions used below are documented in Section 1: Methodology.
The highest score achieved on any model (Claude, Gemini, or ChatGPT), per business, was used as that business's headline score. Across all 298 health practices:
| Statistic | Value |
|---|---|
| Mean (highest score per business) | 14.00 |
| Median (highest score per business) | 12.0 |
| Standard deviation | 4.61 |
| Minimum | 1 |
| Maximum | 50 |
These figures describe the observed sample only. No confidence interval is reported, since the sample was not drawn as a random probability sample of all Miami-Dade health practices — it reflects the commercially sourced contact list described in Section 1.
| Outcome | Meaning | Businesses |
|---|---|---|
| True zero | No relevant signal — category not discussed at all | 0 (0%) |
| Category-adjacent, unnamed | Category and often competitors discussed — business never named | 296 (99.3%) |
| Genuinely named | The business itself was named by at least one model | 2 (0.7%) |
The 99.3% figure is the sharper finding. These health practices aren't falling outside AI's field of view — the opposite is true. ChatGPT, Claude, and Gemini are actively discussing their exact category and, in the large majority of cases, naming their competitors in the same response. The business itself is simply omitted from the answer.
2 businesses held a genuine, direct name citation in this dataset.
| Business | City | Claude | Gemini | ChatGPT |
|---|---|---|---|---|
| Weston Cosmetic Surgery Center | Weston | 1 | 21 | 50 |
| Miami Obstetrics & Gynecology | Miami | 11 | 21 | 50 |
Every one of the 298 health practices tested received at least a category-adjacent response — none returned a complete absence of signal.
Because our detection method captures category-adjacent responses rather than discarding them, we can extract exactly which businesses get named in place of the tested health practice — using only names that actually appeared in the model's response. This is the same entity-signal gap covered in our AI Search Optimization Guide.
Counts represent the number of distinct health practices in this study whose ChatGPT, Claude, or Gemini responses named this organisation at least once. Each business contributes at most one count per organisation, regardless of how many of the three models named it — so counts cannot exceed the 298 health practices tested.
| Business named | Type | Times cited | % of businesses |
|---|---|---|---|
| Cleveland Clinic Florida | Hospital system | 51 | 17.1% |
| Baptist Health South Florida | Hospital system | 47 | 15.8% |
| Jackson Memorial Hospital | Hospital system | 29 | 9.7% |
| Mount Sinai Medical Center | Hospital system | 27 | 9.1% |
| BayCare Physical Therapy | Regional chain | 16 | 5.4% |
| Broward Health | Hospital system | 14 | 4.7% |
| Physiotherapy Associates | National chain | 12 | 4.0% |
| University of Miami Health System | Academic medical center | 12 | 4.0% |
| Jackson Health System | Hospital system | 11 | 3.7% |
| NovaCare Rehabilitation | National chain | 9 | 3.0% |
| Miami Physical Therapy | Regional chain | 9 | 3.0% |
| Nicklaus Children's Hospital | Hospital system | 9 | 3.0% |
| Elite Physical Therapy | Regional chain | 8 | 2.7% |
| Baptist Health Primary Care | Hospital system | 8 | 2.7% |
| Bascom Palmer Eye Institute | Academic medical center | 7 | 2.3% |
Across the sampled responses, AI systems most frequently recommended larger institutions. When an independent health practice isn't named, the fallback is almost never "a different independent competitor" — it's an institution with far deeper third-party coverage.
Where AI names a specific alternative, the qualities it associates with them cluster around a small set of themes:
| Attribute | Times mentioned |
|---|---|
| Personalized care / treatment plans | 71 |
| Comprehensive services / care | 45 |
| Multiple locations | 36 |
| Advanced technology | 12 |
| Patient-centered approach | 10 |
| Laser treatments | 10 |
| Holistic approach | 9 |
| Injectables | 8 |
None of these are differentiators unique to large institutions — these are exactly the kind of claims an independent health practice's own website already makes. The gap isn't in what independent health practices offer; it's in whether AI has independent, third-party confirmation of it.
See Full Methodology → for how these scores were generated.
99.3% of 298 health practices received a category-adjacent response. 2 of 298 achieved a genuine citation (score 35+) on at least one model. 0 of 298 achieved citation on more than one model. Mean highest-model score across the sample was 14.00 (median 12.0, standard deviation 4.61). These are the directly measured results of this study.
The observed citation patterns are consistent with systems that rely heavily on externally confirmed entities drawn from multiple indexed sources, though this study did not have access to any model's internal retrieval process and cannot confirm the underlying mechanism directly. This is an interpretation of the observed findings above, not an additional measurement.
Signal gap vs ranking gap. This does not look like a ranking problem. It looks like an entity visibility problem — these AI systems surface entities with sufficient external confirmation signals, and large institutions simply have vastly more of that confirmation than any independent health practice.
0 of 298 health practices in this dataset achieved citation on more than one model. Within this dataset, any observed citation behaviour appeared to be model-specific rather than consistently reproduced across all three systems.
This health dataset is one of three sectors tested in Miami using identical methodology and detection thresholds.
| Sector | Tested | Category-adjacent | Genuine citation |
|---|---|---|---|
| Health (this report) | 298 | 99.3% | 0.7% |
| Legal | 120 | 100% | 0% |
| Real estate | 97 | 99.0% | 0% |
The pattern replicates closely across all three Miami sectors, despite health, legal, and real estate having nothing in common structurally. See the full Miami AI Search Visibility Study for the complete cross-sector analysis of all 515 businesses.
| Metric | New Jersey (n=216) | Pennsylvania (n=163) | Miami (n=298) |
|---|---|---|---|
| Category-adjacent, unnamed | 98.6% | 96.3% | 99.3% |
| Genuinely named (1+ model) | 0.5% | 2.5% | 0.7% |
| Genuine cross-model citation | 0 | 1 | 0 |
Miami's health results sit close to the earlier New Jersey and Pennsylvania figures, extending the institutional-dominance pattern to a third state. See the Miami AI Visibility Study for the full cross-sector and cross-state analysis.
Health Practices without AI citation presence risk reduced discovery in AI-native search, while their most likely "competitor" in a ChatGPT, Claude, or Gemini answer is a large institution with a permanent structural advantage in third-party coverage.
Third-party citation signals are associated with a higher likelihood of AI citation presence but do not guarantee inclusion in any specific AI-generated answer. AI model outputs are probabilistic and change over time.
For full detail behind these answers, see Full Methodology → and the AEOGeoAI methodology page.
AEOGeoAI.
Miami Health AI Search Visibility Study 2026.
Report ID: AEOGEOAI-MIA-HEALTH-2026-001.
Version 1.0. Published July 2026.
https://aeogeoai.net/miami-health-ai-visibility-study
@report{aeogeoai_miami_health_2026, title = {Miami Health AI Search Visibility Study 2026}, author = {{AEOGeoAI}}, year = {2026}, month = {7}, institution = {AEOGeoAI}, url = {https://aeogeoai.net/miami-health-ai-visibility-study}, note = {Report ID: AEOGEOAI-MIA-HEALTH-2026-001} }
| Version | Date | Changes |
|---|---|---|
| 1.0 | July 2026 | Initial publication |
We publish structured entity articles about Miami-Dade health practices on verified local publications already indexed by AI systems, creating additional third-party entity references that may improve the likelihood of AI citation over time.
View Miami Health service and pricing →From $399 · One-time publication · No retainer · Miami-Dade & Broward Counties