Q&A · 20 questions
Hospital AI search visibility — 20 questions and answers
Published 2026-09-27 · Boily, Korea
The questions clinics in Korea ask most about appearing in ChatGPT, Claude, Gemini and Perplexity answers — measurement, why a clinic does not appear, working with Boily, international patients and compliance. The full Korean set has 76 questions. No visibility or ranking is guaranteed.
Measurement
- Q. How do you measure whether a clinic appears in AI answers?
- A. A fixed set of questions patients actually ask — area, specialty, symptoms, treatments, situations — is put to ChatGPT, Claude, Gemini and Perplexity every two weeks. The mention rate is the share of calls in which the clinic is named; questions containing the clinic's own name are excluded from the denominator. More
- Q. Why biweekly and not weekly?
- A. AI answers vary from run to run, so most weekly change is noise. A two-week interval improves the signal, and three months gives six data points — enough to see a trend.
- Q. Why must the question set stay fixed?
- A. Changing the questions breaks comparison with earlier rounds. A set is frozen at the first measurement; if it is changed later, a new baseline starts.
- Q. Is the API measurement the same as what patients see in the app?
- A. Not exactly. Measurement runs through the AI services' programmatic access with web search enabled to approximate the app, but an individual user's screen can differ. Results are best read as trends on a fixed set, not as a screenshot.
- Q. What is the difference between mention rate and the AI recommendation score?
- A. Mention rate counts whether the clinic is named at all. The recommendation score weights position: per engine, not mentioned 0, mentioned 0.3, near the top 0.7, first 1.0. A clinic can be mentioned often but rarely first. More
Why a clinic does not appear
- Q. Our clinic has a website, reviews and experienced doctors. Why doesn't AI mention us?
- A. Being a good clinic and having material AI can read are different things. The most common pattern is that the facts exist only inside images, scripts or apps, or are inconsistent across channels, so there is nothing the AI can quote.
- Q. Why do we appear on one AI but not another?
- A. Each engine reads different sources. It is common for the same clinic to be high on one engine and zero on another, so the four engines are read separately. More
- Q. Why do the same few clinics always appear?
- A. They are named consistently in the documents the AI reads. Rather than trying to displace them on general questions, the realistic path is to fill questions nobody owns yet — symptoms, specific treatments, situations.
Working with Boily
- Q. What does Boily do?
- A. Boily measures how Korean hospitals appear in four AI engines every two weeks and, if the hospital wants, carries out on-site remediation and re-measures on the same set. It is a measurement and remediation company, not an advertising agency. No visibility or ranking is guaranteed.
- Q. How much does it cost?
- A. KRW 880,000 per month (VAT included) per hospital set (specialty × area), including biweekly measurement and remediation, with a three-month minimum. Payment by bank transfer with an electronic tax invoice. The first measurement is free.
- Q. How long does it take?
- A. Plan on at least three months (six rounds). The first month is usually quiet; clinics starting near zero often take several rounds to move. Rounds that go down are common, so direction is judged on three or more rounds.
- Q. What happens if it doesn't improve in three months?
- A. If the contract contains a performance clause, that clause applies; clauses differ by contract and some contracts have none. Judgement uses the average of recent rounds rather than one round. Boily does not sign contracts that guarantee visibility or rank.
- Q. What happens to our data if we cancel?
- A. Round records remain. The cancelled question set is kept ‘dormant’ and re-measured on the same set every two months, up to three times, so the hospital can see what happens after remediation stops. On renewal the same set resumes.
- Q. Do reports include rounds that went down?
- A. Yes. Every round goes into the hospital's report as measured, including declines. Reporting only the rounds that rose would remove the basis for the next decision.
International patients
- Q. Can you measure how we appear to foreign patients?
- A. Yes — in English, Japanese, Chinese (Simplified) and Thai, with question sets written in the way patients in each language actually ask. Each language is measured as a separate set. More
- Q. Is it enough to translate our Korean website?
- A. Usually not. What foreign patients ask differs from what Korean patients ask, and AI answers in each language draw on different sources. Measurement per language shows where the gaps are.
- Q. Which language is hardest?
- A. It varies by specialty, clinic and language, so there is no general answer. Measure each language first rather than assume.
Compliance
- Q. Is AI visibility work covered by Korea's medical advertising law?
- A. Yes, because it means writing hospital documents patients read. Guarantees, superlatives, comparisons with other providers, testimonials and inducements are restricted. This is not legal advice. More
- Q. Does Boily check content for medical advertising compliance?
- A. Every sentence Boily writes or edits is checked against a lawyer-reviewed checklist before publishing and compared again on the live page afterwards. It is included at no extra cost; it is not legal advice and not the official pre-review. More
- Q. Can we advertise that ‘AI recommends our clinic’?
- A. Not recommended. Using AI answer screenshots in clinic marketing can read as an endorsement claim. Keep measurement results as internal material. This is not legal advice.
Written by Boily. Answers restate Boily's published Korean Q&A. Nothing here is legal advice. Boily guarantees no visibility or ranking.