Three files that decide whether ChatGPT can see your hospital

If you run a practice or a hospital's digital estate, test this before reading further. Open ChatGPT, type your specialty and city, and ask who the patient should see. If your consultants are not in the answer, that is not a ranking problem — it is a legibility problem, and it is fixable in a fortnight.

1. llms.txt

A plain-text file at the root of the domain stating what the practice does, which clinicians work there, what they specialise in and what they have published. A cover letter written for machines rather than patients. It costs an afternoon and removes guesswork a crawler would otherwise do badly.

2. A schema graph that validates

Not a single Organization block. A connected graph: MedicalBusiness for the practice, Physician entities for each consultant with qualifications and registration identifiers, MedicalWebPage on clinical content, and reviewedBy properties tying articles back to the clinician who checked them. Broken JSON-LD is worse than absent JSON-LD, because it fails silently and nobody notices for two years.

3. A sitemap that reflects reality

Most hospital sitemaps are stale, partial, or list URLs that redirect. If a page is not in the sitemap, discovery depends on internal linking alone. If it is not discovered it is not indexed, and if it is not indexed no answer engine will cite it regardless of how good the clinical content is.

What the three do together

llms.txt tells a crawler what you are. The schema tells it how to categorise and trust what it finds. The sitemap ensures it finds everything. None are visible to a patient. All three determine whether a patient hears your name.