We Spent Three Months Measuring the Patient Journey in Health Tourism
We measured how international patients actually decide: response times, channel roles, and where the funnel leaks. Findings from one clinic's Q4 2025 data.
Why I ran this study, what I measured, what came out of it, and what we started doing afterwards.
I ran this study in the last quarter of 2025. It was the field component of my undergraduate thesis, and then it sat in a drawer for a while. In the time since, I’ve had the chance to put some of the findings into practice, which means I can now write about what we did afterwards rather than only what we found. That felt like the right moment to publish it.
One thing up front: this is not an industry report. It is data from a single clinic, a single quarter, a single group of procedures. Please don’t read any of the figures below as a sector benchmark.
Why I decided to measure
Working in health tourism operations, these were the sentences I heard most often. “Instagram doesn’t bring patients.” “They disappear the moment we send a price.” “Leads are down this month.”
All of it intuition. None of it backed by a number. To be fair, neither were my own claims. When four people in the same room believe four different things about patient behaviour, that usually points to a measurement gap rather than a disagreement.
What I wanted to know was simple. Where does an international patient actually travel before reaching us, at which point do they decide, and what do they trust?
What I measured
I combined three sources.
A survey. A structured questionnaire sent to 60 international patients who had already been treated, between one and four weeks after treatment. The response rate was roughly two thirds. The questions focused on the decision process: which channels they used, at which stage, and what earned their trust.
Analytics. Site traffic for the same three-month period, broken down by channel, along with session behaviour and conversion data.
Conversation analysis. A structured read of inbound messaging. Response times, the stages a conversation moves through, and where it breaks off. Coded thematically at the content level, with patient confidentiality preserved.
What I could not measure. This matters, because it sets the boundary around everything else. The data comes from one clinic, covers only aesthetic and cosmetic procedures, and spans three months, so it cannot see seasonality. And the biggest gap: I only surveyed patients who converted. The ones we lost midway are not in this dataset, even though understanding their reasons would probably be the most useful thing of all.
What came out of it
Patients don’t arrive through a single channel
Not one patient in the sample went from first exposure to booking through a single channel. On average, a patient touched several social posts, returned to the website more than once, and opened more than one messaging conversation before deciding, across roughly five to six days.
Three typical journeys emerged. Patients arriving through social saw an image, went to the profile, then the site, then messaging. Patients arriving through search landed on the site, read a blog article and the pricing page, then moved to messaging. Patients arriving through referral ran a quick verification pass and wrote directly, and decided noticeably faster than everyone else.
The part that surprised me: patients from search and referral also checked the social account, usually in the middle of the messaging conversation. So social seems to be doing a second job beyond bringing in leads, closer to a verification layer. When we measure Instagram only by the leads attributed to it, that second job is invisible.
Roughly speaking, the channels had divided the work between them. Social carries awareness and verification, the website carries information and institutional trust, and messaging carries persuasion and closing.
Response speed mattered far more than I expected
This was the finding that surprised me most. Between conversations answered within the first fifteen minutes and those answered after four hours, there was roughly a fivefold difference in conversion. The decline isn’t a single cliff either. It drops at one hour, drops again at four, and a conversation left until the next day is effectively over even when the patient is still replying politely.
There was a side finding too. Late replies didn’t only convert less, they spent more messages getting nowhere. Fast conversations resolved in around four exchanges, while slow ones ran past eight and were still lost.
The reason isn’t hard to guess. The person on the other side is considering surgery abroad, in a language that isn’t theirs, at a clinic they found on their phone. They cannot evaluate the surgeon’s hands. They can evaluate whether you answered.
Follower count sat at the bottom of the trust list
When we asked patients to rate what made them trust the clinic, the top of the list was real before/after imagery from actual patients, followed by the surgeon’s credentials and documented experience. Right behind those came the quality and speed of the messaging itself. Accreditation and video testimonials were close by.
Follower count and engagement numbers sat at the bottom, noticeably below everything else.
The content preference data pointed the same way. Before/after material and process video were rated most useful by a wide margin. “Day in the life of the clinic” content came last, which also happens to be the easiest type to produce.
The funnel leaked somewhere other than where I expected
Mapping the inbound conversations end to end made the shape clear. Attrition through first reply, detailed consultation and sending a quote was moderate. Immediately after the quote went out, roughly half of the remaining volume disappeared.
On the other side of that gap, more than nine in ten patients who paid a deposit completed treatment. Once money moves, the decision is effectively made.
So the most fragile part of the process is the narrow window between the quote and the deposit. At the time, we were spending most of our budget and energy upstream of it.
Nationality shifted the script more than age did
Age changed the channel. Younger patients leaned toward short-form video and decided faster, while older patients went straight to the website, preferred long-form content, and took two to three times longer to decide.
Nationality changed the argument. British patients ran the widest price comparisons and asked the most about legal recourse. German patients wanted technical depth and documentation, and vague answers cost trust with them. Middle Eastern patients decided fastest, but expected personal attention and communication in their own language.
What we started doing afterwards
The real value of the study began here. Rather than leaving the findings in a report, we focused on a few things in the period that followed.
We made response time visible. The first step was simply putting the number in front of people. Who picked up which conversation and when, and how long untouched conversations were waiting. It became clear early that hiring more consultants wasn’t going to solve it, because the underlying issue is coverage across five time zones. That needs ownership rules, routing, and escalation for untouched conversations. Less a discipline problem than an infrastructure one.
We began building a separate flow for the post-quote stage. The data pointed at that window as the narrowest part of the funnel, and we had no structured process there. What happened to a patient who went quiet after seeing the price was left to the consultant’s judgement.
We shifted the content balance toward evidence. More consented before/after material and process video, less filler. This one moves slowly, because it depends on patient consent and real documentation.
We changed what we report. We stopped presenting follower growth as if it were a business metric, and replaced it with response time and stage-by-stage conversion.
We started separating the script by market. Instead of a single “international patient” approach, one that accounts for which objection arrives first in which market.
These changes worked. Our response times came down, we lost fewer patients after sending a quote, and the team stopped arguing about what to prioritise. That was the study’s real return: it turned a process run on instinct into one we could measure.
One closing note
This data comes from a single clinic. If your procedure mix, patient profile or market breakdown differs, your results may well differ too, so I’d compare these figures against your own before carrying any of them across.
The method travels better than the numbers do. You don’t need an academic study to measure your response times, your stage-by-stage conversion, or the route a patient actually takes to reach you. Reading the data you already hold, once and properly, gets you most of the way there.
If you’ve run something similar at your own clinic, I’d like to compare notes. This field has plenty of confident statements and very few shared numbers.
This article draws on the field findings of my 2025 undergraduate thesis at Istanbul University, on the influence of digital communication channels on international patients’ decision-making in health tourism. Figures have been generalized to protect clinic and patient confidentiality.