Home / Resources / Healthcare AI Visibility Study

We asked ChatGPT about 20 Indian hospital chains. 13 were invisible.

When someone asks AI "which hospital in Mumbai is best for cardiac surgery?" or "recommend a good maternity hospital in Bengaluru," they're making a decision that matters far more than which brand of earphone to buy. Our study found that 13 of India's 20 leading hospital chains were entirely absent from those conversations — and 5 of the 7 that did appear carried errors that could actively mislead a patient.

How we tested

We selected 20 hospital chains and networks with significant national or regional presence — including Apollo Hospitals, Fortis Healthcare, Narayana Health, Max Healthcare, Aster DM Healthcare, Manipal Hospitals, Medanta, HCG Hospitals, Yatharth Hospital, Shalby Hospitals, Cloudnine, Kokilaben Dhirubhai Ambani Hospital, Wockhardt Hospitals, Ruby Hall Clinic, Lilavati Hospital, Hinduja Hospital, Motherhood Hospitals, Sakra World Hospital, NH network facilities, and NH-affiliated specialty centres.

We ran three prompt types across ChatGPT (GPT-4o), Gemini Advanced, and Perplexity:

• Specialty prompts — "best cardiac hospital in India", "top oncology hospital in Bengaluru", "which hospital is recommended for knee replacement in Delhi?"
• Direct prompts — "what is [hospital chain] known for?", "is [hospital chain] good for [specialty]?"
• Comparison prompts — "[hospital A] vs [hospital B] for [specialty] — which would you recommend?"

Visibility required an unprompted appearance in specialty queries. Accuracy required no material error on specialties, city presence, or clinical programs.

What we found

• 13 of 20 hospital chains never appeared in any specialty query across any engine, despite collectively treating millions of patients annually.
• Of the 7 that did appear, 5 carried material errors — wrong specialty strengths cited, individual flagship hospitals presented as the entire network, cities listed where the chain has no presence, and outdated information about acquisitions and rebrandings.
• Only 2 of 20 were both visible and accurately described — a 10% pass rate in a category where patients are making consequential health decisions based on AI answers.
• AI engines consistently defaulted to citing government institutions (AIIMS Delhi) for specialty queries, regardless of whether the question asked about private networks — suggesting private chains have almost no AI footprint relative to the information vacuum around them.

What the invisible 13 had in common

1. Clinical expertise locked in PDFs and print brochures. Every chain has deep clinical capability — but that expertise lives in downloadable brochures, printed patient guides, and internal SOPs. None of it is in a format AI retrieval systems can read, quote, and cite in an answer.
2. Practo and JustDial ratings don't feed AI answers. Most hospital chains invested heavily in review management on aggregator platforms. Those platforms aren't in AI training pipelines the same way that structured web content, Wikipedia citations, and indexed news coverage are.
3. Individual hospitals overshadow the network. A flagship Kokilaben or Ruby Hall has strong individual identity — but the chain entity that owns them is often poorly defined in the AI's knowledge base. AI doesn't know how to navigate the difference between the brand and the unit.
4. No owned original health content. The chains with AI visibility published original clinical content — patient guides, outcome data, treatment explainers — that gave AI something authoritative to reference. Invisible chains relied entirely on their OPD booking pages.

What to do about it

Healthcare brands can build AI presence without regulatory risk: publish structured clinical content, document specialty programs in crawlable HTML, build entity presence in Wikipedia and health knowledge bases, and earn citations in health journalism. The GEO playbook for healthcare doesn't require patient data or clinical claims — it requires clear, accurate, machine-legible descriptions of what you treat and how. Full method in our Complete GEO Guide.