Himedi

What I did

• I led design for the U.S. storefront, from the homepage that makes the case for booking through Himedi to the screens where a customer places an order.
• I built the brand system, from the typefaces and the palette down to the kind of photograph we would use, because nobody at the company had one to hand me.
• I took my designs to production by directing AI coding tools, since there was no front-end engineer on the team, so what a customer saw was what I had designed.

Who it’s for

Himedi began as a concierge service that had to convince Americans a checkup in Korea was worth the trip, so the site asked people to decide whether to travel and what to book in the same visit. We eventually narrowed the audience to people who already had a trip to Korea planned, and for them the only question left was why they should book through us rather than go straight to a clinic.

I approached it by mapping what that person already knew against what they still had to be shown, and that turned the site from a pitch for Korea into a set of answers in the order a visitor would ask for them. The quiz became the door for anyone who arrived without a checkup in mind, so the pages that mattered could stay simple.

The homepage

The homepage had one job, which was to convince someone who already had a trip to Korea planned that booking through us was the better way to get a checkup there. So I decided it would make that one argument and moved the quiz and the tier comparison onto separate pages, where a visitor who was ready to choose could take their time.

I wrote the argument down as three plain reasons before I designed anything, and then built the page to carry them. The price would be the clinic’s or less, the package would be built around them, and the appointment would be booked, translated and handled for them.

The booking flow

The flow had to keep the promise the homepage made, that the checkup would be built around whoever was booking one. So a customer picked a base checkup, added the areas they wanted covered, then reviewed and booked, and what they ended up with was one they had put together themselves.

The first version sold three tiers of depth, which read like an upsell once I went through it as a customer, so I broke that one choice into two smaller questions. I had the flow work out price and eligibility itself rather than have someone set them by hand on each product, since there were far too many combinations to maintain one at a time.

Building it

There was no front-end engineer on the team, so the only way my designs reached production was for me to direct AI coding tools to build them. That let me put a layout into the real product in an afternoon and go through it the way a customer would, on a phone, which is where most of Himedi’s visitors arrive. The problems a mockup would never have shown me turned up while there was still time to fix them, and what shipped was the whole path from the homepage’s argument to a booked checkup.

The brand system

The brand was mine to make, because nobody at Himedi handed me a palette or a typeface when I arrived. Every value a visitor saw on the site, down to the colour of a button, was there because I had decided it should be.

I cut the site’s type down to a short scale of sizes and two faces, one for headlines and prices and one for reading. That way anything a customer read, from a hero headline to a price in the cart, was set the same way on every page.

The three focus-area colours were chosen to be light and friendly, which made them fail as button fills once white text sat on them. So I made a deeper version of each one for the buttons and kept the light ones for borders and icons.