aeogeoai.netMiami AI Search Specialist
Blog →
AEOGeoAI Research is an independent research programme investigating how generative AI systems discover, validate and recommend businesses. Published work includes Research Report No. 001 — the New Jersey AI Search Visibility Study 2026 (216 independent health practices tested across ChatGPT, Claude and Gemini; 98% showed zero AI citation presence; DOI: 10.5281/zenodo.20949937) — and Research Report No. 002 — the Miami Health Practices & Google AI Visibility Report 2026 (DOI: 10.5281/zenodo.20918793). Both studies are deposited on Academia.edu and Zenodo. AEOGeoAI Research is based in Miami Beach, Florida. Research themes: AI search visibility, AI citation behaviour, entity recognition, local AI search, Google AI Overviews, Google AI Mode, ChatGPT, Gemini, Claude, AI brand visibility, GEO, AEO. Contact: [email protected]
Original research into AI search visibility, AI citation behaviour and entity recognition. AEOGeoAI publishes research examining how businesses and organizations are represented across ChatGPT, Google AI Mode, Google AI Overviews, Gemini, Claude and other AI search systems.
Our work combines original datasets, AI visibility testing, entity analysis and local market research to understand how generative AI systems discover, validate and recommend businesses.
The first multi-model AI citation visibility study of its kind. 216 independent New Jersey health practices tested across ChatGPT, Claude and Gemini. Results established the baseline evidence problem facing independent local businesses in AI search.
Analysis of AI search visibility for independent health practices in Miami-Dade. Proposes a two-stage entity retrieval model describing how generative AI systems validate and geographically associate local businesses before citing them in recommendation answers.
500 independent Miami businesses across three verticals — health practices, law firms and real estate developers. Cross-model AI citation analysis, source attribution, geographic entity signal testing and cross-model comparison. Expected Q3 2026.
The following factors are associated with AI citation visibility based on AEOGeoAI's published research findings. Association does not imply causation — these factors appear consistently among businesses with measurable AI visibility but have not been tested under controlled conditions. All findings are specific to local business recommendation queries.
The single strongest observed correlate. Businesses cited by generative AI systems consistently have structured coverage in independent indexed publications. In the NJ study, the four practices with any AI visibility all had independent editorial coverage; 98% with zero coverage had zero AI visibility across all three models.
Miami example: A Brickell immigration law firm cited by South Florida Reporter is far more likely to appear in ChatGPT answers about Miami immigration attorneys than an uncovered firm with an identical website.Coverage in high-domain-authority national publications appears to establish entity existence at a level generative AI systems recognize independently of local signals. AEOGeoAI's own MSN.com placement produced Position 1 on Google and an AI Overview citation within seven days of publication.
Miami example: A Coral Gables cosmetic surgery practice featured on MSN.com signals verified entity existence to ChatGPT and Gemini at a national level — separate from any Miami-specific local signal.A separate signal from entity validation. generative AI systems appear to require location-specific confirmation from locally indexed sources before including a business in geographically-qualified recommendation answers such as "best dermatologist in Coral Gables." Both stages — national entity validation and local geographic association — appear necessary simultaneously.
Miami example: A Wynwood medspa cited by a Miami-Dade publication satisfies the geographic association requirement. National coverage alone, without local confirmation, may not be sufficient for geographically-qualified queries.Consistent name, address, category and credential information across all independent sources. Inconsistent entity information — different business name formats, different address formats, different category descriptions — creates conflicting signals that appear to reduce AI confidence in citing a business. This applies across publications, directories, GBP, schema markup and review platforms simultaneously.
Miami example: A Miami Beach real estate developer listed as "Brickell Development Group," "Brickell Dev Group LLC," and "BDG Miami" across different sources creates entity ambiguity that may suppress AI citation even where coverage exists.FAQ schema creates extractable question-and-answer units that generative AI systems can lift directly into generated answers. This is a structural content signal distinct from third-party authority — it improves the extractability of existing content rather than establishing external entity evidence. Most effective when combined with primary publication coverage.
Miami example: A Miami immigration law firm with FAQ schema answering "What is the EB-5 visa process for Venezuelan investors?" provides an extractable unit ChatGPT can incorporate directly into a recommendation answer.The specificity and consistency of geographic entity signals across all indexed content. Includes: zip codes (33139, 33131, 33134) in content and schema; 305/786 area codes in consistently formatted contact information; GPS coordinates in image metadata and LocalBusiness schema; Wikidata geographic identifiers for neighborhoods and locations; geotagged images with coordinate metadata; and entity-rich noun density in GBP review content and owner responses.
Geographic precision appears to help generative AI systems disambiguate entities in competitive local markets. "Miami Beach" is ambiguous across dozens of businesses; "33139" and GPS coordinates at a specific address are not.
Miami example: A luxury condo development on Lincoln Road with geotagged images described as "Preconstruction condo development, Lincoln Road, South Beach, Miami 33139" — with GPS coordinates embedded in image EXIF data — provides machine-readable geographic specificity beyond what address text alone conveys. Wikidata entry Q201516 for Miami Beach further grounds the entity in a structured knowledge graph generative AI systems read.Review text containing specific entity nouns — practice name, specialty, procedure names, doctor names, location — contributes to indexed entity association signals in ways that generic sentiment ("great service!") does not. Review volume alone is a weak signal; review content specificity appears more relevant to AI entity recognition. Owner responses that mirror the same entity-rich noun set reinforce the signal from both sides of the indexed conversation.
Miami example: A Coral Gables plastic surgeon whose GBP reviews mention "Dr. [Name]," "rhinoplasty," "Coral Gables," and "international patients" in the review body — and whose owner responses repeat the same terms — produces indexed content richer in entity-specific nouns than a 5-star review reading "amazing experience."Structured content formatted specifically for AI crawler consumption, including llms.txt files and AI-targeted summary pages. An emerging practice with limited independent verification. Low-cost to implement and consistent with the direction of AI crawler development. Listed here as an observed emerging signal rather than a proven citation factor.
Miami example: An llms.txt file at aeogeoai.net/llms.txt providing structured descriptions of Miami AI search services and research assets provides generative AI systems an explicit, machine-readable entity summary without requiring page rendering.Wikipedia presence correlates with AI citation visibility but is a lagging indicator of citation density, not a cause of it. Businesses with Wikipedia pages had typically already accumulated sufficient independent third-party coverage across local publications, directories and editorial sources to make community-created documentation likely. Pursuing a Wikipedia page without first establishing the underlying citation footprint is working backwards — and Wikipedia pages created without genuine notability are removed.
Miami example: A long-established Aventura healthcare group with decades of local news coverage, multiple directory citations, and consistent entity information may eventually acquire a Wikipedia page as a downstream effect of that coverage — not because anyone created the page strategically.Presence on Yelp, Zocdoc, Avvo, Healthgrades and similar platforms appears in AI responses as category-awareness context but rarely as confident entity-citation sources for specific local recommendation answers. These platforms contribute to general entity awareness rather than the specific third-party validation required for confident AI citation in recommendation queries.
Schema markup helps generative AI systems parse on-site content but does not substitute for independent third-party evidence. A business with complete LocalBusiness schema and zero external citations remains invisible to AI recommendation systems. Schema is a necessary supporting signal, not a primary citation driver.
Different generative AI systems weight citation factors differently. ChatGPT favours structured editorial sources. Gemini pulls heavily from Google-indexed properties and GBP signals. Claude responds well to evidence-rich, cited content with clear authorship. Cross-model visibility — appearing in ChatGPT, Gemini and Claude simultaneously — requires coverage across multiple source types rather than optimizing for a single platform. In our NJ study, no practice achieved cross-model visibility; the four with any visibility appeared in only one model each.
Miami example: A Miami real estate developer cited in MSN.com (strong Gemini signal) and South Florida Reporter (strong ChatGPT signal) is better positioned for cross-model visibility than a developer with deep coverage in a single source type.How businesses appear within AI-generated answers across ChatGPT, Google AI Mode, Google AI Overviews, Gemini and Claude — and how that presence differs from traditional search ranking.
How generative AI systems select, reference and attribute external sources when generating recommendation answers — including which publication types and content formats produce reliable citation signals.
How AI models identify, validate and disambiguate businesses, locations and brands across independently indexed sources — and what consistency, specificity and coverage signals influence that process.
How geographic signals — zip codes, area codes, local publications, GPS coordinates, Wikidata geographic identifiers — influence AI recommendation answers for location-qualified queries.
The practice of building external entity evidence in third-party publications that generative AI systems cross-reference when generating answers. Distinct from traditional SEO, which targets Google's ranking algorithm rather than AI citation systems.
Structuring on-site content — FAQ schema, direct-answer formatting, structured Q&A — so generative AI systems can extract and cite it confidently in generated answers. Complements but does not replace GEO.
Research findings are published regardless of commercial implication. AEOGeoAI Research does not adjust findings to favour client outcomes or service positioning.
All testing parameters — prompts, models, scoring, date of testing — are documented and published. Findings are not reported without the methodology required to evaluate them.
Studies are designed to be reproducible by independent researchers using the published methodology. Probabilistic variance in AI outputs is acknowledged and accounted for in scoring.
Reports are dated and versioned. Where findings are updated — due to AI model changes, additional data, or methodological improvements — prior versions remain accessible and differences are documented.
Published studies are registered with a Digital Object Identifier (DOI) via Zenodo, providing permanent, citable references independent of AEOGeoAI's own infrastructure.
Every AEOGeoAI research report follows a consistent methodology covering prompt construction, scoring, AI model selection, probabilistic output handling, date stamping, repeatability limitations and sampling approach. A permanent methodology reference is maintained separately so individual reports can be read without methodology repetition.
When referencing AEOGeoAI published studies, please use the following formats:
AEOGeoAI Research Team (2026). New Jersey AI Search Visibility Study 2026. AEOGeoAI. Miami Beach, Florida. DOI: 10.5281/zenodo.20949937
AEOGeoAI Research Team (2026). Miami Health Practices & Google AI Visibility Report 2026. AEOGeoAI. Miami Beach, Florida. DOI: 10.5281/zenodo.20918793
AEOGeoAI Research operates independently while informing the practical consulting services provided by AEOGeoAI. Research questions arise from real-world observations, while consulting recommendations are grounded in published findings wherever possible.
Research is based in Miami Beach, Florida. Current focus: Miami-Dade local markets across health, legal and real estate verticals, with particular attention to the bilingual English/Spanish AI search environment that makes Miami structurally distinct from other US markets.
Contact: [email protected] · Miami Beach (305)-709-0437