We read 122 English-language reviews of eight Seoul skin and aesthetic clinics. Not one of them said how long the result lasted. Not one mentioned a consent form. Not one mentioned anaesthesia, a refund, or the doctor’s specialty. What these reviews leave out is more informative than what they contain — and there is a documented reason for it.
This is the first in a series in which we treat clinic reviews as data rather than as opinion. We do not host reviews and we do not rank clinics. We count what is there, we count what is not, and we publish the base every time.
What we did
On 7 August 2026 we collected every English-language review visible to us on Google Maps for eight clinics in Gangnam, Apgujeong and Hongdae, taking both the default sort order and the lowest-rating sort order. That produced 122 reviews: 52 at five stars, 13 at four, three at three, one at two, and 53 at one.
It is not a random sample and we will not pretend otherwise. Google stops paginating after roughly eight to ten reviews per sort order — one clinic in our set displays 1,506 reviews and showed us eight. Of the 122 we did read, 102 were truncated by Google at around 240 characters, so anything mentioned in the tail of a review is invisible to us. Every count below is a floor, not a measurement.
What was missing
| What we looked for | Reviews mentioning it |
|---|---|
| How long the result lasted | 0 |
| A consent form or any paperwork | 0 |
| Anaesthesia | 0 |
| A prepaid package or course of treatments | 0 |
| A refund, or compensation of any kind | 0 |
| The doctor’s board certification or specialty | 0 |
| Legal recourse, insurance, or a complaint body | 0 |
| Aftercare or a follow-up visit | 1 |
| A written quote before payment | 1 |
| An actual price in Korean won | 2 |
The zeros are the finding. Lifting, fillers and botulinum toxin were the procedures most often named in this corpus, and all three are bought on the promise of a duration. Not one reviewer said whether they got it.
What was there instead
If reviewers were not describing outcomes, what were they describing? We classified the 65 positive reviews the same way.
| What the positive reviews praised | Count of 65 |
|---|---|
| The premises — cleanliness, decor, “luxury” feel | 29 |
| Staff warmth and friendliness | 20 |
| A clear explanation of the procedure | 11 |
| A named individual member of staff | 11 |
| Satisfaction with the result | 8 |
| Transparent pricing | 6 |
| An interpreter or English-speaking staff | 3 |
| Not being pushed to buy more | 2 |
Twenty-nine reviewers praised the room. Eight praised the outcome. Complications appear four times in the negative reviews and never once in the positive ones, which is what you would expect if the positive reviews were written before any complication could have appeared.
The axis of evaluation has moved. These are not reviews of medical care. They are reviews of an afternoon.
“They didn’t push me” is a compliment
Thirteen of the 57 negative reviews describe being sold additional treatments — in two cases, being moved into a treatment room without being told what was about to happen. Against that, two positive reviews single out the absence of selling as the thing worth writing down.
They did not try to upsell me at all, just listened to my needs and recommended the best…
Google review, Reberry Gangnam, 5 stars, July 2026
The pricing is honest and transparent — no hidden fees or pushy upsells.
Google review, Cellin Hongdae, 5 stars, April 2026
A market in which not being sold to is remarkable enough to record is a market in which being sold to is the default.
Who does the selling is its own finding. Across the corpus, the person conducting the consultation is called a consultant, a counsellor, a receptionist, a “foreign manager” or, flatly, “sales people”. Only two reviews state how long the patient spent with a doctor.
I spent over two hours with counselor but I had maximum three minutes with Dr. Lee.
Google review, Banobagi, 1 star
To start with I had only 5 min consultation then they rushed me out to serve next person.
Google review, ID Hospital, 1 star
Why the reviews stop at the door
Two reviews, written by different people about different clinics, describe the same mechanism.
They give you free gifts in exchange for a 5 star review. They ask you to post it during the consultation. Then they check the review before giving your gift. So beware the ratings may be inflated.
Google review, MUSE Gangnam, 1 star, June 2026
The high review you see here is because they ask you to gove a five star review before you leave, so you get a gift. You have to show that you have a five star review and that you posted it.
Google review, Reberry Gangnam, 1 star, July 2026
Neither is a copy of the other. The phrasing differs, the spelling differs, and one contains a typo the other does not. Both describe the same four steps: the request arrives during or before the procedure, the rating is specified, the posting is verified, and only then is the gift handed over.
If a review is written inside the clinic on the day of treatment, it can only ever describe that day. That is the entire explanation for the first table in this article. There is no duration in these reviews because there was no duration yet.
A regulator has already documented this
Two reviews are two reviews. But in July 2026 Korea’s Fair Trade Commission issued corrective orders against three Seoul plastic-surgery clinics for the same practice, and the published account of the case describes the mechanics in more detail than any patient could.
The clinics recruited patients as promotional models through their own websites, discounted the surgery in exchange, and then sent instructions by messenger: a specified minimum character count, before-and-after photographs required, and a post once a month for a year after surgery. At least one required a deposit of 500,000 won, returned on compliance.
The finding worth carrying is not the sanction. It is the principle. A review written by a real patient about a real operation is still deceptive advertising if it was compensated and the compensation was not disclosed. Authenticity is not the test. Disclosure is.
No fine was imposed — the outcome was a corrective order. For comparison, the equivalent United States rule, in force since October 2024, carries a maximum civil penalty above fifty thousand dollars per violation.
The only two prices in 122 reviews
Exactly two reviews in the corpus contain a figure in Korean won. One of them is a four-star review, and the reviewer is not complaining.
The price in store was 88,000 won if you bring your Korean ARC card. If you don’t have one they’ll charge you 97,000.
Google review, MUSE Gangnam, 4 stars, June 2026
An alien registration card is what a foreign resident of Korea carries. A visitor does not have one. The reviewer records a ten per cent difference in the same tone she uses for the rest of the visit — as a fact about the day, not as a grievance. That is the corpus in miniature: things are recorded, not evaluated.
The second figure comes from a one-star review and puts the gap far higher.
If you call a Korean call center they will tell you a completely different price it will be at least 30% cheaper than their foreign manager tells you, so be careful
Google review, VIEW Plastic Surgery, 1 star
The two figures do not agree with each other, and we are not going to pretend that two reviews establish a pattern. What we can say is that we could find no systematic measurement of what foreign patients are quoted at Korean clinics — not by a regulator, not by a researcher, not by anyone. An entire industry has been built on English-language search without that number ever being published.
What this does not prove
- It does not establish that any particular clinic solicited reviews. It establishes that two patients say they were asked to, and that a regulator found three other clinics doing it.
- It does not tell us what proportion of reviews are produced this way. Google shows eight per sort order; we cannot see the rest.
- A five-star-heavy distribution is not by itself evidence of manipulation. Naturally occurring review distributions are already skewed, and self-selection alone produces skew. We treat that signal as weak and will deal with it properly in a separate article.
- Silence is not the same as absence. A reviewer who does not mention a consent form may well have signed one.
What we could not confirm
- Korean-language reviews. Naver Place, KakaoMap, Gangnam Unni and Babitalk were all unreachable to us. Everything above describes the English-language corpus only, and the comparison with Korean patients that we plan for a later article rests on a much smaller community sample.
- The Fair Trade Commission decision itself. We worked from published accounts of the case, not the decision document. Reported dates for the announcement differ by a few days across outlets, which is why we date it only to July 2026.
- Whether complaints skew towards English. At one clinic, the ten lowest-rated reviews were all in English. That is suggestive and nothing more; we did not read the Korean reviews.
- Whether foreign patients are systematically quoted more. Two reviews in this corpus mention a gap and disagree about its size — ten per cent in one, thirty in the other. Two data points are not a finding, and we could locate no study that measures it.
Check a clinic yourself
Every signal used in this article can be checked by a reader in about two minutes: count the star breakdown you can actually see, open ten reviews and count how many come from accounts with a single review, and note whether you were offered anything in exchange for posting. We built a page that does the arithmetic and shows you what each signal is worth and where it comes from. It runs in your browser and stores nothing.
One thing has become easier very recently. Since 9 July 2026, Naver publishes each reviewer’s own average star rating — so an account whose every review is five stars is now visible to anyone looking.
Collected and verified 7 August 2026. Figures are counts of what we read, not estimates. If you find an error in this article, write to [email protected] with the specific claim and we will correct it on this page with the date and a note describing what changed.