How to Get Your Clinic Cited by ChatGPT, Perplexity, and Gemini (AEO Guide for Singapore Healthcare)

Why are ChatGPT, Perplexity, and Gemini now a distribution channel for clinics?

In 2024, if a Singapore patient wanted to find a dentist in Orchard, they typed “dentist Orchard” into Google and clicked the map pack. In 2026, a meaningful slice of those same patients ask ChatGPT “what's the best-rated dentist in Orchard that does Invisalign and takes evening appointments?”. ChatGPT gives a single answer. If your clinic isn't in that answer, you're invisible to that patient — whether or not you rank on google.com.sg.

This is not hypothetical. ChatGPT has 400 million weekly active users globally. Perplexity hit 20 million monthly users in 2025. Gemini is embedded directly inside Google Search as AI Overviews, which appears above the classic ten blue links for the majority of healthcare queries. The channel is already significant and is growing 4-5x year-over-year in APAC.

How do AI engines actually decide which clinic to cite?

The decision process is different from classical Google ranking. AI engines look for:

• Extractable answers. A page that says “The average cost of Invisalign in Singapore is SGD $4,500 to $8,000 depending on complexity” is citable. A page that says “Invisalign costs vary” is not. AI engines need a self-contained answer they can quote verbatim.
• Question-formatted H2s. Pages where most H2 headers are phrased as the question a patient would ask (“How much does Invisalign cost in Singapore?”) are cited roughly 3x more often than pages with statement H2s.
• Verifiable citations. AI engines weight sources that cite named research (MOH, Singapore Dental Council, HSA, Pew Research) higher than sources that assert claims without attribution.
• Structured data. Speakable, FAQPage, MedicalClinic, and LocalBusiness schema help AI crawlers identify citable segments of the page.
• Consistency signals. Name, address, phone, opening hours across Google Business Profile, your website, Healthhub, and third-party directories. Inconsistency tells AI engines the data isn't reliable, and they quietly drop the clinic from recommendations.
• Review velocity. Same signal as Google local pack. 80 recent reviews at 4.8 stars beats 400 old reviews.

What does the AEO structural checklist look like in practice?

Here's the minimum viable AEO structure for a Singapore healthcare page:

1. H1 with the primary query verbatim
2. Opening paragraph that directly answers the query in 2-3 sentences
3. 3-6 question-formatted H2s, each followed by a 40-80 word direct answer
4. Stat callout with a specific number and named source
5. FAQ section with 3-5 Q&A pairs, each Q&A self-contained
6. Sources block with citations to named authorities (MOH, Pew, Google)
7. JSON-LD: Article + FAQPage + Speakable + MedicalClinic
8. Internal links to 2-3 related pieces on your site

This is the format AI engines reliably cite. It's also the format patients read comfortably. AEO and good content-design point in the same direction — unlike the old SEO tricks of keyword-stuffing and thin pages.

What is llms.txt and how do I ship one?

llms.txt is a plain-text file at the root of your domain that tells AI crawlers what to cite and in what order. It's the AEO equivalent of robots.txt. A minimum viable llms.txt lists your clinic name, your primary service, 3-5 priority pages, and a short positioning summary in 150 words or less. Ship it once, maintain it quarterly, and you're ahead of 95% of Singapore healthcare sites.

OpenAI, Anthropic, and Perplexity have all publicly committed to respecting llms.txt once the spec finalises. Google hasn't committed yet but is observing. Shipping the file now costs nothing and compounds over the next 12-18 months as AI engines tighten their citation rules. If you'd rather have your site scored against all of the above first, run the free 220-point clinic audit — it scans AEO structure, schema, llms.txt, and citation-readiness in under 60 seconds.

How do I know if any of this is working?

By running weekly AI-citation probes and logging the answers. For each target query you care about (“best aesthetic clinic Paragon”, “24-hour vet Clementi”, “TCM acupuncture Chinatown”), query ChatGPT, Perplexity, and Google AI Overviews, and log the full transcript. If your clinic is cited, great — you have evidence. If it's not, the transcript tells you who else is, which is a direct competitive intelligence signal.

Most Singapore agencies don't do this. They pitch AEO at the concept level, can't show you a single screenshot of a client being cited, and charge a vague “AEO package” premium. If you're evaluating an AEO provider, ask to see actual probe transcripts for an existing client. If they can't produce them, they don't have AEO infrastructure — they have an invoice line item. We publish our own probe log in the Logara self-rank case study — same methodology, same engines, with screenshots.