Hotel Calm
Revisited

Returning to an early project with more experience to make it stronger, and more honest.

A booking app, revisited.

Role

Solo UX & UI Designer

Scope

Mobile app · End-to-end flow

Timeline

2023 · Revisited 2026

Tools

Figma · Miro

Hotel Calm is a fictional boutique hotel chain with locations across Europe: Stockholm, Athens, Barcelona, Lisbon, Marseille and Paris. The mobile app is the primary booking channel for guests who already know the brand.

This case study is a return. I designed Hotel Calm in 2023, early in my UX education at the UX Design Institute, producing both a UX version (research, flows, wireframes) and a UI version (design system, screens, prototype).

The research found the right problems. The design didn’t fully solve them. Room cards carried almost no information, prices sat on top of photographs, and checkout gave no sense of how much was left. Revisiting it in 2026, I wanted to close the gap between what the research said and what the interface did.

The original UX process, as I did it the first time.

Research

Usability tests with six users across two existing hotel booking sites, three users per site, plus benchmarking of Airbnb, Trivago, Agoda, Kayak, Design Hotels and others. The findings that mattered most: users missed the breakfast option, lacked information to compare rooms, and wanted to check the total before paying.

SIX SESSIONS ACROSS TWO BOOKING SITES

Affinity diagram

Feedback clustered into filter, calendar, info & rating, pictures, map, add-ons and booking. The clearest signals: users missed the breakfast option, lacked room information and wanted to double-check the total before paying.

THIS IS WHERE THE BREAKFAST PROBLEM FIRST APPEARS

Customer journey

Seven stages from Ready to Booking. Confidence, as I read it across the six sessions, dropped during search and filtering, then rose sharply once users saw strong imagery.

CONFIDENCE DROPS HERE, DURING SEARCH AND FILTERING

Flow diagram

The happy path for two guests booking a room with breakfast, in three stages: search, room and add-ons, checkout.

NINE SCREENS FROM START TO CONFIRMATION

STAGE 1Location, dates, guests

STAGE 2Room and add-ons

STAGE 3Checkout and confirmation

KEY · ROUNDED RECTANGLE = SCREEN · CIRCLE = USER INTERACTION

Wireframes

Start screen, location picker, content pages and the booking flow, connected as a clickable prototype.

THE FULL FLOW, CONNECTED AS A CLICKABLE PROTOTYPE

A warm, editorial look for a fictional hotel.

Colour

A warm, muted palette inspired by Field Notes notebooks. Tactile and editorial, it allows imagery to remain the primary focus.

Black
#121212

Dark brown
#442211

Beige
#DCD6C5

Light beige
#E8E6DB

Dark yellow
#B58C24

Yellow
#E9B420

Red
#9B2B15

Light red
#DC6243

Gray
#D5D5D5

White
#FFFFFF

UI elements

Inter throughout, set directly on imagery as often as on a plain ground. The effect is striking, but legibility varies with whatever photograph sits behind it.

Photography

Front-facing perspective, centered composition, warm colour treatment. Natural light and considered interiors, chosen to carry the hotel's character before any text is read.

Looking back at the first design.

What worked

  • The overall flow was complete, from start screen to booking confirmation
  • The confirmation section concept was genuinely good. Showing choices in real time is the right instinct
  • The BOOK button was large, yellow and clear
  • The final "See you soon" screen had the right tone
  • The visual identity was strong throughout

What didn't work

The confirmation section

The original concept was to keep booking choices visible throughout the flow, so guests could review their selections as they progressed. The panel showed location, dates and guests, but not the selected room or the total price. It solved part of the problem the research found, without giving guests a complete picture of their booking.

The original research had already found most of these problems. The first design noticed them, but didn't fully solve them.

Learning from other booking experiences.

The Hoxton

Same segment · Boutique, multi-city

  • Room names with personality: Shoebox, Snug, Cosy, Biggy
  • Spec row: size, bed type, max guests
  • Total price per stay including tax
  • Persistent "Your Stay" panel with running total
  • Key lesson: personality and specifics can sit in the same card.

Mollie's

Budget boutique · brand-first

  • Content and feeling before the booking
  • Value proposition stated directly on the homepage: rooms from £70, bar, WiFi, in-room touchscreen
  • FAQs on the start page eliminate doubt early
  • Consistent local tone: "Hello Oxford", "It's a Mollie's thing"
  • Key lesson: concrete details do more than adjectives.

Booking.com

Aggregator · The opposite of boutique

  • Overwhelming information density
  • Everything on one long checkout page
  • No brand feeling
  • Key lesson: more information does not equal more confidence

Four lessons for the redesign.

01

Name rooms, don't describe them

"Shoebox" communicates size, feel and use in one word. "Small room" communicates nothing.

02

Show the total, including tax

The Hoxton shows it at every step. The 2023 flow showed a nightly rate and revealed the total only at payment.

03

Show how many steps remain

The 2023 checkout never said how far along you were. "Step 2 of 3" is one line of text that fixes it.

04

Give add-ons context

A price and a sentence change the decision. "Breakfast · served 7 to 10am · €15/night" is a different offer than the plain dropdown I built.

Five design decisions, each traced back to what the original research found.

Room cards redesigned

"User lacking information on the different room options, despite different pricing"
FROM THE AFFINITY DIAGRAM

The 2023 cards gave a price and "Fitting 2 adults". Nothing to compare on. Each room now has a name, a spec row and a short description. The price sits on a solid surface below the image instead of over the photograph, where it was legible only by luck.

Before
After

The Your Stay panel

"User wants to double check the total booking before final purchase"
FROM THE AFFINITY DIAGRAM

The 2023 panel listed your choices but never what they cost. Your Stay shows room, dates, nights, guests, add-ons and a running total, updating from the moment a room is selected, so the total is visible before the payment screen rather than after it. Prices are in euros across all six cities.

Before
After

Add-ons get their own screen

"User misses to book breakfast, did not see call to action button"
FROM THE AFFINITY DIAGRAM

In 2023 the add-ons were three items in a dropdown, and the research showed users missing breakfast entirely. They now get their own step, each with a price and a line of description. That adds a step to the flow, but a step that gets noticed is worth more than a step saved.

Before
After

Checkout split into three steps

The 2023 checkout was one screen: nine fields, personal details and card details in a single scroll, with no sign of how far along you were. I split it into three labelled steps and gave every field a label that stays visible. In the original, "Oliver" and "Oliver Flygare" sat four fields apart with nothing to say which was the cardholder.

FROM AFFINITY DIAGRAM: "Too many or too detailed choices at the last part of booking journey"

Before
After

Final confirmation screen

The tone here already worked, so I kept it. What changed is the information: the full Your Stay summary carries through, and "Email will be sent to you" became "A confirmation has been sent to Oliver@gmail.com". Past tense, with the address, so nothing is left pending.

Before
After

A self-initiated project with no live data, so these are the measures I would track.

Booking completion rate

The share of users who start a booking and finish it.

Drop-off in checkout

Where people leave checkout, and at which of the three steps.

Error rate

How often users go back to change a room, date or add-on before paying.

Add-on adoption rate

The share of completed bookings that include at least one add-on.

Booking clarity

Whether guests can state their room, dates, add-ons and total before paying. Measured in usability testing.

Consistent rules for typography, spacing, colour and action.

From start to confirmation.

The complete redesigned booking flow is available as an interactive Figma prototype, from start screen to booking confirmation.

OPEN PROTOTYPE →

What the revisit taught me.

What I learned

The first version of Hotel Calm was designed from the outside in. I started with how it should look, then filled in the flow. Revisiting it showed me that the order should be reversed: understand the problem, decide what the user needs to know at each step, and let the visual design support that.

Three lessons stand out.

Copy is part of the design. "Small room / info" wasn't a placeholder problem, it was a thinking problem. Writing real room names and descriptions forced me to decide what actually matters to a guest.

Good ideas need complete execution. The confirmation panel was the strongest idea in the original, but it only confirmed half the booking. A half-built idea can create more doubt than no idea at all.

Competitors are a design resource, not just a benchmark. The Hoxton showed me what good looks like in the same segment. Booking.com showed me what to avoid. Mollie's showed me something I haven't acted on yet: that trust starts before the booking does.

What I'd do next

Test the redesign. The v2 is based on insights and competitor patterns, not user testing. My next step would be a round of usability testing with 5 users, measuring task completion time and where they hesitate in checkout.

Run an accessibility review. Text placed on imagery was a recurring pattern in the original. I'd check contrast, touch target sizes and screen reader flow across the full journey.

Design the before and after. Mollie's showed that trust is built before the booking starts. I'd explore how content and brand pages could lead users into the booking flow, and what happens after it: check-in, in-stay services, and the journey home.

Three years between the two versions, and the research didn't change. Only what I did with it.