Why most businesses score zero in AI search — and it has nothing to do with their website
In June 2026 we ran the first multi-model AI citation visibility study of its kind, testing 216 independent health practices in New Jersey across ChatGPT, Claude and Gemini. The result: 98% showed no AI citation presence across any of the three models. Zero. Not low — absent entirely.
These were real practices — many with fully optimised websites, Google Business Profiles, strong local SEO, and years of established reputation. None of that mattered. AI systems didn't cite them because AI systems don't read their websites when generating local recommendation answers.
The core insight: AI systems like ChatGPT, Google AI Overviews and Gemini generate local recommendations by cross-referencing independent third-party publications — not business websites. If you have no presence in those publications, you do not exist in AI-generated answers. Full stop.
The same pattern applies across industries and locations. We subsequently observed it across Miami-Dade health practices, legal services, and real estate. The mechanism is identical regardless of market. Most businesses have an evidence gap — and they don't know it.
SEO vs AI search optimization — why they're not the same thing
Understanding the difference between traditional SEO and AI search optimization (GEO — Generative Engine Optimization) is the prerequisite for everything else in this guide. They target different systems, optimise different signals, and produce different results. A business that excels at one can score zero on the other.
Optimises your website for Google's ranking algorithm
Targets page position in blue-link results. Primary signals: backlinks, on-page keyword relevance, site authority, technical health. Success metric: ranking position for target keywords. Does not influence what AI systems say when generating recommendation answers.
Builds external entity evidence AI systems cross-reference
Targets citation presence in AI-generated answers. Primary signals: third-party publication placement, consistent entity information across independent sources, geographic confirmation. Success metric: named in AI-generated local answers. Completely independent of website ranking.
A dentist in Coral Gables can rank #1 on Google for "cosmetic dentist Coral Gables" and score zero across ChatGPT, Google AI Mode and Gemini simultaneously. The patient who asks Google AI Mode for a recommendation never sees them — even though they rank first in traditional results. These are genuinely separate channels.
Traditional SEO is still necessary. The point is not to abandon it. The point is that it does not, on its own, produce AI citation presence.
How AI systems actually decide which businesses to cite
When a patient in Brickell asks ChatGPT "find me an immigration attorney in Miami," or a buyer in Bogotá asks Google AI "best new construction condos in Miami Beach 2026," the AI doesn't search Google. It cross-references its indexed knowledge of third-party sources — publications it has already learned to treat as credible, independent validators of local entities.
The decision to cite a specific business follows what we describe as a two-stage entity retrieval process:
This is why so many businesses are invisible in AI search despite strong Google rankings. They satisfy neither stage — because their entity evidence lives entirely on their own website, which AI systems don't treat as independent third-party validation.
How to rank in AI search results — the actual mechanism
Given the two-stage model, the path to AI search visibility is straightforward in principle and specific in practice. You need structured entity evidence in independent publications AI systems already treat as trusted sources. The publication choice matters as much as the content.
Step 1 — National authority publication (Entity Validation)
A high-domain-authority national publication gives AI systems the entity validation signal. MSN.com is the anchor in our publication framework. An article on MSN.com tells ChatGPT, Google AI and Gemini that a named entity has been independently covered by a nationally recognised editorial source. That is a different signal from anything your website can produce.
An MSN.com article published through our network ranked Position 1 on Google organically and was cited as a source in Google's AI Overview within seven days of publication. The AI Mode tab was visible in navigation — confirming the article was being pulled into AI-generated answers.
Step 2 — Local publication (Geographic Association)
National authority alone is not enough. AI systems also need geographic confirmation — a local publication in the same market that corroborates the entity's location and service category. For Miami-Dade, that means verified local publications that Google AI and ChatGPT already treat as credible sources for local entity validation. The format of the article matters: data-heavy factual articles, answer-first profiles, and FAQ-structured pieces each produce different AI extraction signals.
Step 3 — Consistent entity information across sources
The same name, address, category, and credentials must appear consistently across every source. Inconsistency in entity information — different business name formats, different address formats, different category descriptions — creates noise that makes AI systems less confident when generating recommendations. Consistency builds the cross-referenced signal.
The combination that works: MSN.com (national authority) + verified local publication (geographic association) = two-layer AI citation signal that covers both stages of entity retrieval simultaneously.
What does not work on its own
How to optimize for AI search — the content format
The publication choice determines whether the article reaches the right AI systems. The article format determines whether AI can extract a confident citation from it. Both matter. An article on the right publication in the wrong format produces a weaker signal than the same article written for how that publication's AI extraction actually works.
| Format | Best for | AI extraction signal |
|---|---|---|
| Data-heavy entity article | Health practices, law firms | Factual confirmation — name, credentials, address, specialty, years established |
| Answer-first profile | Premium services, consultants | Direct answer format — exactly what AI Overviews prefer to pull |
| FAQ-structured article | Local businesses, service practices | Multiple extractable answer units — strong for both AI Overviews and ChatGPT |
| National news press release | All business types via MSN.com | Authority signal — entity validated by national editorial source |
Every article should include the business's exact name (consistent with all other sources), full address, specific service category, lead person's credentials, and a direct answer to the most likely AI query in the first paragraph. AI systems extract the opening of an article preferentially. The answer needs to be there, not buried three paragraphs in.
AI search performance — what to measure
AI search performance is measured differently from traditional SEO. Ranking position is not the metric. Citation presence across models is.
| Metric | How to measure |
|---|---|
| AI citation presence | Query your business name + category + location in ChatGPT, Google AI Mode and Gemini. Note which models cite you and for which queries. Use aeogeoai.net to score across models. |
| Cross-model visibility | Whether you appear in more than one AI system. In our NJ study, zero practices achieved cross-model visibility. Two-layer publication placement is required. |
| Query coverage | Which category + location queries return your business. Expand coverage by adding publication placements for additional query types. |
| AI-referred traffic | In GA4, filter by referrer for chatgpt.com, perplexity.ai, and gemini.google.com. AI-referred visitors convert 4.4× better than traditional search traffic (Semrush, 2026). |
AI outputs are probabilistic and vary across sessions. A single query test is not definitive — test across multiple sessions and all three models. Use aeogeoai.net for a structured score across ChatGPT, Claude and Gemini.
What this looks like in practice — Miami examples
The principles above are market-agnostic. The Miami-Dade market is where we've done the deepest implementation work — which makes it the clearest illustration of the mechanism in action.
A Miami immigration attorney
When a prospective client in Caracas asks ChatGPT "best immigration attorney in Miami for Venezuelans," AI generates a direct answer naming a firm. Before our publication framework, that firm would typically be one with press coverage in national legal publications — not the best attorney in Miami, but the one with the best third-party evidence footprint. After a two-layer placement (MSN.com + South Florida Reporter), a firm without prior AI presence begins appearing in ChatGPT responses for immigration attorney queries within weeks.
A cosmetic practice in Coral Gables
Our NJ study found that 98% of independent health practices are absent from AI search entirely. The same is true in Miami. A cosmetic dermatology practice with a fully-optimised website, 200+ Google reviews, and a 4.9 rating can score zero across ChatGPT, Gemini and Google AI Mode — because none of those signals are third-party entity evidence AI systems use for recommendation answers. One structured article on the right Miami-Dade publication changes that.
A real estate developer in Miami Beach
When an investor in Bogotá asks Google AI Mode "best new construction condos Miami Beach 2026," Google AI does not return a portal. It names a development. That development appears there because it has entity-rich structured coverage in sources Google AI cross-references. For Miami Beach real estate, LatAm investor queries are the high-value channel — and most developments have zero AI citation presence in it.
The mechanism is identical across verticals. The publication stack changes by market. The principle — third-party entity evidence at national + local level — does not.
Is your business invisible in AI search?
Check your AI visibility score free across ChatGPT, Claude and Gemini — then see our done-for-you Miami placement service if you want the fix handled for you.
Check your AI visibility score →10 free checks · No sign-up · Miami service from $299 one-time · (305)-709-0437