GEO / AI-Search Audit — Miami Vitality & Aesthetics

This is the dimension where your clinic has the most ground to make up — and, frankly, the most upside. You have built an aggressive, modern marketing machine that converts the people who already know about you. But a growing share of patient discovery no longer happens on the open web you optimized for. It happens inside AI answer engines — Google's AI Overviews, ChatGPT, Perplexity, Gemini, Microsoft Copilot — that read the web, decide which businesses are trustworthy and well-described, and then answer the patient directly without sending them to ten blue links.

GEO (Generative Engine Optimization) is the discipline of making your clinic legible, citable, and quotable to those systems. Your 38 reflects a site built for human persuasion and conventional SEO, not for machine comprehension. The good news: the same fixes that make you more citable to AI also make your offers clearer and your compliance posture safer — three birds, one stone. The catch: a chunk of your aggressive health-claim copy is exactly the kind of language AI engines are now trained to distrust and route around, which means your marketing strength is quietly undermining your AI visibility.

This audit walks through where you stand on six GEO sub-dimensions, how each of the five major AI platforms currently "sees" you for the searches that matter, what's blocking the crawlers, and a prioritized plan to climb.

A GEO score is not a search ranking and we will never frame it as one. AI engines do not publish rankings, do not guarantee placement, and change their selection logic frequently. What we measure is readiness — the set of structural, content, and authority signals that make it more likely an AI system can find you, understand you correctly, and cite you confidently when a patient asks a relevant question. Think of it as preparing your clinic to be a good source, not gaming a leaderboard.

Scores map roughly to: 80–100 AI-native (consistently surfaced and cited); 60–79 competitive (surfaced for some queries, inconsistent); 40–59 partially visible (occasionally referenced, often via third parties); below 40 largely invisible (AI engines either skip you or describe you from directory data you don't control). At 38, Miami Vitality sits just under that line — present in the ecosystem, but mostly being described by others rather than quoted from your own site.

The pattern is consistent: where the work is about being known locally (brand authority, technical performance) you're respectable; where the work is about being understood and trusted by a machine reading your words (citability, E-E-A-T, schema, platform tuning) you're well below par. That's a fixable, content-and-structure problem — not a reputation problem.

Citability measures how easily an AI engine can lift a clean, factual, self-contained statement from your pages and reproduce it with confidence. AI systems prefer passages that are specific, attributable, and low-risk to repeat. Your site is the opposite by design: it's written to move a reader emotionally toward booking, not to inform a machine precisely.

What's holding you back

• Claim-density over fact-density. Service pages lead with outcomes ("reverse your biological age," "feel 20 years younger," "optimize every system in your body") rather than describable facts ("a 45-minute IV infusion of vitamin C, B-complex, and magnesium administered by a registered nurse under physician supervision"). AI engines can quote the second kind safely; they actively avoid the first, because unsubstantiated superlatives are a trained risk signal.
• No extractable answer blocks. When a patient asks "how much does hormone optimization cost in Miami" or "what is peptide therapy," the engine wants a crisp 40–60 word answer it can attribute to you. Your pages bury this under hero animations, testimonials, and CTAs. There is no clean, quotable definition or price-range block anywhere on the site.
• Pricing opacity. "Pricing varies — book a consultation" is invisible to AI. Engines cannot cite a number you won't state, so they pull ranges from third-party directories and competitor comparisons instead — meaning the AI's answer about your pricing is sourced from people who aren't you.
• Vague authorship. Passages aren't tied to a named, credentialed clinician, so even genuinely good educational content reads as anonymous marketing.

Why this connects to your biggest risk. The same superlative, disease-implying language that creates your FDA/FTC exposure (covered in depth in the Compliance module) is also the language AI engines are tuned to distrust and skip. Rewriting toward structure/function-style, substantiated, plainly-worded claims is simultaneously a compliance fix and a citability fix. This is the single highest-leverage change in the entire GEO program.

Target state. Every service page opens with a 2–3 sentence factual definition, carries a clear price range or starting-from figure, attributes educational claims to a named clinician, and includes a short FAQ written as direct question-and-answer pairs.

Brand authority measures how strongly the wider web — the corpus AI models actually learn from — recognizes Miami Vitality as a real, established entity. This is your strongest GEO sub-dimension, and it reflects the genuine reach of your marketing.

Working in your favor

• ~260 Google reviews at 4.4★ is a substantial, authentic signal of a real, busy clinic. AI engines weight review volume and recency heavily when deciding whether an entity is "real enough" to mention.
• An aggressive, active social presence (Instagram, TikTok, paid social) keeps the brand name circulating, which strengthens entity recognition.
• Consistent NAP (name, address, phone) across the major local directories, from what we can see, helps engines resolve "Miami Vitality & Aesthetics" to a single coherent business.

The gap

• Authority is shallow and self-published. Your footprint is overwhelmingly owned and social — your site, your Instagram, your ads. AI engines lean hardest on third-party, editorially independent sources: local press, healthcare directories with editorial standards, reputable wellness publications, physician-profile sites, and structured knowledge bases. Miami Vitality is thin here.
• No knowledge-panel-grade presence. There is no Wikipedia/Wikidata entry (appropriate for your size — not a recommendation to force one), and the physician's independent professional profiles are sparse or inconsistent. When an engine wants to verify who runs the clinic, it struggles.
• Medical-vertical directories under-leveraged. For a medspa/longevity clinic, profiles on healthcare and aesthetics directories (the platforms AI engines treat as semi-authoritative for medical entities) are incomplete or unclaimed.

A 55 here means AI engines believe you exist and matter locally — they just don't have enough independent corroboration to cite you confidently over a competitor with better third-party coverage.

E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is the framework AI engines and search systems use to gauge whether content about health topics is safe to surface. Health is "Your Money or Your Life" territory — the bar is deliberately high. Your content is built to persuade, which is a different (and sometimes opposing) goal.

Dimension by dimension

• Experience (weak). Real patient experience is present as testimonials but rarely as described clinical experience — "our physician has administered over 4,000 hormone optimization protocols since 2016" type statements that demonstrate hands-on depth. Testimonials alone read as marketing, not experience.
• Expertise (weak–moderate). The clinic is physician-led, which is a genuine asset — but the site under-uses it. There's no robust clinician bio with credentials (medical license, board certification, training, affiliations), no author bylines on educational content, and no clear "medically reviewed by" attribution.
• Authoritativeness (weak). Almost no outbound citations to authoritative sources (peer-reviewed literature, FDA, professional medical societies). Health content that asserts mechanisms and outcomes without referencing anything is a trust red flag for both AI engines and regulators.
• Trustworthiness (weak). Missing or buried: required FDA-style health-claim disclaimers ("These statements have not been evaluated by the Food and Drug Administration…"), clear notes where products/uses are not FDA-approved, FTC-aligned "results not typical" context on testimonials, a transparent pricing posture, and an easily found, specific privacy policy describing how patient data is handled (HIPAA expectations). Their absence hurts trust scoring and compliance at the same time.

Why it scores 35. The clinic has real expertise and real experience — they're just invisible in the content. E-E-A-T isn't about having authority; it's about demonstrating it on the page in a way both humans and machines can verify. Right now the marketing voice is crowding all of it out.

Technical GEO covers whether AI crawlers can physically reach, render, and extract your content. Your site is modern and fast for humans — but "fast and modern" sometimes works against machine extraction.

Strengths

• Good Core Web Vitals and mobile performance; HTTPS everywhere; clean, marketing-friendly URL structure.
• Fast initial load and a polished mobile experience — genuinely good human UX.

Problems specific to AI extraction

• JavaScript-dependent rendering. Key content (service descriptions, pricing context, FAQs) appears to load client-side via the site builder's framework. Some AI crawlers execute JavaScript poorly or not at all; if the meaningful text isn't in the server-rendered HTML, those engines see a near-empty page. This is likely the single biggest technical drag on your AI visibility.
• Content trapped in images and embeds. Important information (price lists, protocol details, before/after framing) is frequently rendered as graphics or inside third-party widgets, which is invisible to text-based extraction.
• Thin, flat information architecture. Heavy reliance on long single-scroll landing pages rather than discrete, well-titled service pages makes it hard for an engine to map "this clinic offers X, Y, Z" cleanly.
• Heading structure optimized for visual design, not semantics. Multiple <h1>s, decorative headings, and skipped levels make the document outline ambiguous to a parser.

Quick technical wins (detailed in recommendations): ensure server-side rendering or static pre-rendering of core content; move pricing and protocol facts into real HTML text; clean up the heading hierarchy; add an XML sitemap that lists every service page individually.

Schema (structured data, in JSON-LD) is how you tell an engine, in its own language, "this is a medical clinic, here is its name, address, the services it offers, its hours, its reviews, and its physician." It removes guesswork. AI engines lean on schema heavily to build their internal model of an entity. Yours is minimal.

Current state (composite assessment)

• Basic LocalBusiness markup may be present via the site builder, but no MedicalClinic / MedicalBusiness typing, which is the correct, more specific type for your vertical.
• No Physician / Person schema tying the named clinician to the practice with credentials.
• No MedicalProcedure / Service / Offer schema describing IV therapy, hormone optimization, peptides, or injectables individually.
• No FAQPage schema, despite FAQs being one of the most directly AI-citable structured formats available.
• No AggregateRating / Review schema exposing your 4.4★ / ~260-review strength to engines in a machine-readable way — you have the reputation asset but aren't declaring it structurally.

Important compliance note. Schema must be truthful and substantiated. We will not mark up disease-treatment claims as MedicalProcedure indications where that would assert a therapeutic outcome the clinic can't substantiate — that would turn a structured-data win into a regulatory liability. Schema gets built in lockstep with the claims cleanup.

A 30 reflects "the site builder added the bare minimum automatically, and nothing vertical-specific or service-specific exists." This is one of the fastest sub-scores to move, because schema is additive code that doesn't require redesigning the site.

This measures how well your content is tuned for the specific ways each AI answer engine assembles a response — especially for the two query types that drive your business: "near me" local queries ("best IV therapy near me," "hormone clinic Miami," "medspa near Brickell") and service/education queries ("does peptide therapy work," "Botox vs filler difference," "what is NAD+ IV therapy").

Right now your content isn't written in the shapes these engines reward — direct answers, comparison tables, clearly-scoped local pages, and FAQ pairs. The next section breaks this down platform by platform.

A reminder before the table: none of these engines publish rankings or guarantee inclusion, and all of them change behavior frequently. What follows is our best read of how each system currently tends to handle queries in your category, and where Miami Vitality stands. We optimize for readiness, not for promises.

Google AI Overviews

For a "near me" medspa query in Miami, AI Overviews leans heavily on Google Business Profile, Maps proximity, review volume/rating, and structured data. Your ~260 reviews at 4.4★ are an asset here — but without MedicalClinic and FAQPage schema, and with service details locked in JavaScript/images, the Overview is more likely to summarize you from your GBP listing and third-party directories than to pull nuance from your site. Highest-ROI platform for you, because your review strength gives you a head start and the gaps are addressable.

ChatGPT (search mode)

ChatGPT's web search favors text it can read server-side and corroborate against independent sources. Your JavaScript rendering plus thin third-party authority is a double penalty: it may not see your real content, and even when it does, it lacks outside corroboration to cite you over a competitor. For "what is NAD+ therapy / where in Miami" style queries, expect it today to describe the category and cite directories rather than Miami Vitality specifically.

Perplexity

Perplexity is the most citation-hungry engine — it shows its sources inline. That makes it the best opportunity for a clinic that publishes genuinely quotable, specific content, and the harshest critic of vague claim-copy. With clean factual definitions, stated price ranges, and FAQ pairs, you become directly citable. Today, with little to safely quote, Perplexity routes around you.

Gemini

Deeply tied to the Google ecosystem, Gemini rewards a consistent, well-described local entity across GBP, Maps, and the web. Your local signals are decent, so readiness is moderate; the unlock is entity consistency (identical NAP, claimed and complete profiles, schema) so Gemini resolves "Miami Vitality" cleanly and confidently.

Bing Copilot

Copilot draws on the Bing index and structured data. Many cash-pay clinics neglect Bing Places entirely; claiming and fully optimizing it, plus adding schema and real HTML text, is low-effort and lifts both Copilot and ChatGPT (which shares Bing-style sourcing).

Before any of the above can work, the engines have to be allowed in and given a map. Two files govern this.

robots.txt — current read (composite). Your robots.txt appears to be the site-builder default. It likely doesn't block the major AI crawlers — but it almost certainly doesn't deliberately welcome them either, and we found no explicit handling of the newer AI user-agents. Crawlers to verify and consciously allow (assuming you want maximum AI visibility, which for a discovery-hungry clinic you do):

• GPTBot (OpenAI training) and OAI-SearchBot (ChatGPT search) — allow
• Google-Extended (controls Gemini/AI Overviews training use) — allow
• PerplexityBot — allow
• ClaudeBot / anthropic-ai — allow
• Bingbot (powers Bing + Copilot + ChatGPT search retrieval) — ensure allowed
• CCBot (Common Crawl, feeds many models) — allow

The risk here isn't a hard block; it's passive neglect — defaults that leave money on the table and no affirmative signal that you want to be read.

llms.txt — currently absent. llms.txt is an emerging convention: a single Markdown file at your root that gives AI systems a clean, curated map of your most important pages and facts, free of marketing clutter and navigation noise. It's not yet a ranking factor and we won't pretend it is — but it's a low-cost, forward-looking signal that makes your key content easy for engines to find and understand. You don't have one. Adding a well-built one is a quick win.

Example llms.txt outline for Miami Vitality

# Miami Vitality & Aesthetics

> Physician-led medical spa and longevity-wellness clinic in Miami, FL.
> Services: IV nutrient therapy, hormone optimization, peptide therapy,
> Botox and dermal fillers. Cash-pay. Physician-supervised.

## About
- [About the clinic & medical team](https://example.com/about): physician
  credentials, supervision model, facility, year established.
- [Our physician](https://example.com/team/dr-lastname): license, board
  certification, training, areas of focus.

## Services
- [IV nutrient therapy](https://example.com/iv-therapy): what it is,
  protocols offered, who administers it, starting price, session length.
- [Hormone optimization](https://example.com/hormone-optimization):
  description, eligibility, supervision, price range, FDA-status notes.
- [Peptide therapy](https://example.com/peptides): description, scope,
  required disclaimers, price range.
- [Botox & dermal fillers](https://example.com/injectables): treatments,
  injector credentials, pricing, expected results context.

## Pricing
- [Pricing & packages](https://example.com/pricing): starting-from figures
  and ranges by service.

## Patient information
- [FAQ](https://example.com/faq): direct Q&A — safety, cost, what to expect.
- [New patient info & consultation](https://example.com/consultation).
- [Privacy practices / HIPAA notice](https://example.com/privacy).

## Required notices
- Health claims have not been evaluated by the FDA; services are not
  intended to diagnose, treat, cure, or prevent any disease, except where
  a specific FDA-cleared/approved use is stated.
- Individual results vary; testimonials are not typical outcomes.

This file does two jobs at once: it hands engines a clean map, and it surfaces your required compliance notices in a structured, machine-readable place — turning a regulatory obligation into a trust signal.

Critical

• Claim-heavy, unsubstantiated marketing copy is both a compliance liability and a citability killer. AI engines distrust and route around superlative health claims; regulators (FDA/FTC) treat them as exposure. This is the root issue dragging citability (30) and E-E-A-T (35) down together. (See the Compliance module for the full regulatory treatment — it is the audit's biggest finding overall.)
• Core content is JavaScript/image-dependent and may be invisible to AI crawlers. If engines can't read your service and pricing text server-side, every other GEO effort is built on sand.

High

• No vertical or service-specific schema (MedicalClinic, Physician, MedicalProcedure/Service, FAQPage, Review) — engines are guessing what you are.
• Pricing opacity removes the single most-requested fact from AI's reach, ceding the narrative about your cost to third parties.
• Thin third-party / independent authority — engines can't corroborate you over better-documented competitors.
• Missing required disclaimers and a clear HIPAA-grade privacy notice — a trust and compliance gap.

Medium

• No llms.txt; robots.txt doesn't affirmatively welcome AI crawlers.
• Long single-scroll pages instead of discrete, well-titled service pages — weak information architecture for entity mapping.
• No FAQ content written as direct Q&A — leaving the most AI-citable format unused.
• Heading hierarchy optimized for visuals, not semantics.

Low

• Bing Places likely unclaimed/under-optimized.
• Inconsistent physician profiles across independent professional directories.

• Publish a built-out llms.txt at the root (outline above). A day of work; an immediate forward-looking signal.
• Add FAQPage + MedicalClinic + AggregateRating schema. Additive JSON-LD, no redesign. Surfaces your review strength and clinic identity to every engine at once.
• Confirm and consciously open robots.txt to the major AI crawlers (GPTBot, OAI-SearchBot, Google-Extended, PerplexityBot, ClaudeBot, CCBot, Bingbot).
• Add a stated price range or "starting from" figure to each service. Reclaims the pricing narrative from directories.
• Claim and complete Bing Places — lifts Copilot and ChatGPT-search readiness cheaply.
• Write a real clinician bio page with credentials, license, training, and affiliations.

Phase 1 — Foundation & risk reduction (weeks 1–4). Quick wins above, plus begin the claims-and-disclaimer rewrite hand-in-hand with the Compliance module. This phase simultaneously lowers your FDA/FTC exposure and raises citability — do it first because everything downstream depends on having clean, quotable, substantiated copy.

Phase 2 — Technical extractability (weeks 3–8). Ensure server-side/static rendering of all core content; move pricing and protocol facts out of images into real HTML; fix heading hierarchy; split single-scroll pages into discrete, well-titled service pages; add a per-service XML sitemap.

Phase 3 — Content & E-E-A-T depth (weeks 6–14). Rewrite every service page to open with a factual definition + answer block; add medically-reviewed bylines and outbound citations to authoritative sources; build out a substantive FAQ in direct Q&A form; demonstrate experience with specific, substantiated practice facts.

Phase 4 — Authority & platform tuning (weeks 10–20). Build independent third-party presence (local press, vertical healthcare directories, consistent physician profiles); fully optimize Google Business Profile and Bing Places; tune content shapes (comparisons, local-scoped pages) for each engine; then re-audit to measure the delta.

Expect movement to be gradual and uneven across platforms — AI engines update on their own schedules, and we measure progress by re-audit deltas, never by promised placements. Realistically, the citability, schema, and E-E-A-T sub-scores should move first and fastest because they're under your direct control; authority compounds more slowly.

Methodology: GEO score (0–100) is a weighted composite of six sub-dimensions — AI citability, brand authority, content E-E-A-T, technical GEO, schema markup, and platform optimization — assessed against current AI-answer-engine behavior, AI-crawler accessibility, and structured-data readiness. Readiness signals only; not a ranking, placement guarantee, or legal/regulatory determination. Illustrative fictional composite for demonstration.

← Back to the full report