UX Optimization

UX Optimization

UX Optimization

Data-Driven Design

Data-Driven Design

Data-Driven Design

Conversion

Conversion

Conversion

Continuous Discovery

Continuous Discovery

Continuous Discovery

Alternative Date Suggestions

AI Style Recommendations

Reducing drop-off by helping users find better dates.

AI Style Recommendations

ROLE

Product designer

COMPANY

trivago

INDUSTRY

Travel

YEAR

2023

ROLE

Product designer

COMPANY

trivago

INDUSTRY

Travel

YEAR

2023

ROLE

Product designer

COMPANY

trivago

INDUSTRY

Travel

YEAR

2023

| OVERVIEW

On Trivago, users often bounce when their selected hotel isn’t available, especially if it’s tied to fixed dates. They hit a dead end and leave.

We saw an opportunity to shift that behavior by suggesting nearby alternatives: same hotel, different dates. The goal was to keep users exploring without disrupting their intent.

By surfacing these suggestions directly in the search results, we aimed to reduce bounce, boost engagement, and turn dead ends into new opportunities.

Expected business outcome:
Increase visits to book over time by first reducing bounce rate and improving engagement during hotel searches.

| USER PROBLEM & DISCOVERY

Product intelligence data

High rate of unavailability

37% of users searching for a specific stay land on a hotel that’s unavailable.

Date flexibility is common

38% of those users try different dates through the calendar when faced with unavailability.

Alternative accommodations are explored

31% switch to a different hotel from the alternative list when their first choice isn’t available.

Unavailability causes drop-off

24% bounce the moment they realize their selected hotel is unavailable.

User support data

Calendar interaction feels frustrating

Users often turn to the calendar after seeing an unavailable hotel, but finding open dates means trial and error. They keep selecting ranges only to hit more unavailability, with no clear path to success.

Booking intent remains high

User support messages show users still want the original property, they just can’t easily find when it’s available.

User support ticket quotes

“I want to get a price for a particular holiday unit. I have put in over 15 date ranges and they are all saying not available. I am wanting to know if this property is still available to rent and how I go about looking at this and booking.”

“I have been trying to book a condo on Panama Beach Florida, for next March. However any date I put in says dates not available! I have changed dates and condos too many times to count. Are all these condos all full? Or am I trying to book too early?”

| OPPORTUNITY

By surfacing available alternative dates when a hotel is unavailable, we can reduce user frustration and help them stay on track to book.

Many users are flexible with dates, but struggle to find what’s available. Instead of guessing in the calendar or bouncing, they need clear next steps that keep them engaged and moving forward.

| IDEATION & SOLUTION DEFINITION

With my cross-functional team we mapped out opportunities and brainstormed possible solutions using the Opportunity Solution Tree.

After evaluating feasibility, desirability, and assumptions, we chose to move forward with a simple yet effective concept: suggesting alternative available dates to help users quickly identify when a searched accommodation can be booked.

How might we...
suggest alternative dates for item search users that land on an unavailable date?

| FEASIBILITY & DESIGN EXPLORATION

As we explored solutions, we focused on designing lean but effective options. Early questions shaped our direction:

  1. Could we show prices with the date suggestions?

  2. Which dates were meaningful for the user?

  3. Would users understand and trust the suggestions?

We sketched variations, ran click and usability tests, and collaborated closely with engineers and stakeholders. The goal: reduce complexity and deliver something meaningful, fast.

The solution we launched wasn’t perfect, but it solved the core problem and was the fastest path to impact.


| DEPENDENCIES & CONSTRAINTS

01

Backend dependency

Suggested dates were dependent on other teams’ priorities and infrastructure.

02

No prices available

We learned users wanted to see price context, but this wasn’t technically feasible at the time.

03

Cached data only

Our MVP relied on recent availability data, not real-time checks. It helped, but didn’t fully solve the pain.

04

Shifting design system

We built while the design system was evolving, which meant constant rapid iteration and adaptability.

| FINAL DESIGN LOGIC

We designed for flexibility, the module adapts based on how many relevant dates we could surface.

+3 suggestions: Horizontal scroll
2 suggestions: Full-width buttons
1 suggestion: Single full-width button
0 suggestions: No display (no valid cached dates)

The solution we launched wasn’t perfect, but it solved the core problem and was the fastest path to impact.


We designed for flexibility, the module adapts based on how many relevant dates we could surface.

+3 suggestions:
Horizontal scroll

2 suggestions:
Full-width buttons

1 suggestion:
Single full-width button

0 suggestions:
No display (no valid cached dates)

The solution we launched wasn’t perfect, but it solved the core problem and was the fastest path to impact.

| OUTCOME OF A/B TEST

Increased date selection after seeing unavailable accommodations

Reduced bounce rate across tested journeys

Increased clickouts to booking partners

Users successfully used the date suggestions to find available stays, validating both the concept and its usability. The test was accepted and merged into the live product.

| NEXT STEPS

While the MVP proved effective, we identified key UX opportunities to evolve the solution:

  1. Surface prices alongside suggested dates
    Research showed this would increase clarity and decision-making.

  2. Refine the backend logic
    Ensure suggested dates appear for every unavailable item and not just cached.

| KEY LEARNINGS

Users need guidance, not options

Just showing a calendar wasn’t enough. Curated suggestions removed friction and boosted interaction.

Data gaps ≠ no value

Even cached data helped users move forward. Imperfect info > no info, when expectations are managed.

Product constraints can sharpen focus

Technical limitations forced us to prioritize the highest-impact moments and ship faster.