Case study · GetYourGuide

Improving destination discovery.

Helping travelers in the research phase of planning a trip — from an initial concept that failed in A/B testing to an iterative, research-driven solution that finally moved every success metric.

Duration July 2020 — May 2021

Company GetYourGuide

Team UX manager (me), UX research, Product & content design

Method Concept testing, iterative experimentation

GetYourGuide is one of the world's largest travel-experience marketplaces, connecting travelers with activities run by local operators. Its mission is to "give the world access to unforgettable travel experiences".

While it was very successful at capturing high-intent users from search engines, the performance earlier in the traveler journey was poor. In mid 2020, we kicked off a project to improve the product for travelers who knew their destination but not yet their itinerary of activities. This would allow us to play a bigger part in the travel planning space.

GetYourGuide's target audience is the "Modern Explorer".

A traveler who goes on vacation frequently (4–6 times a year) to experience new cultures. Travel is part of their identity; they take pride in it as a hobby. They want both the tourist spots and the off-the-beaten-path.

Based on visitor data, we identified three levels of intent.

High intent

High-intent customer

Travelers who have already decided on the experiences they want to do. Most likely have booked their flights and hotel.

Mid intent

Mid-intent customer

Travelers who know their destination but are still researching experiences they want to do. Most likely have booked their flights and hotel.

Low intent

Low-intent customer

Travelers who are still deciding on their destination. No flights or hotels have been booked.

When landing on our website, users didn't get a good understanding of how to experience a city.
Users found it difficult to scan and understand the activities on offer.
There was no easy way to conveniently narrow down choices based on their needs and preferences.

Screenshot of the GetYourGuide city page at the start of the project — City landing pages at the start of the project

Problem statement

"Help mid-intent customers understand what makes their destination special, and discover experiences that match their interest and needs."

Many solution ideas were floating around the company. The main challenge was synthesizing organizational knowledge and customer data into a cohesive first concept — both to align scope internally and to get direction-level feedback from users.

Ideas to be tested

We built a cohesive first concept that combined three ideas: a short city intro with key information, top sights to anchor travelers to a destination, and interest-based collections to help narrow down choices.

Methodology for the testing was a RITE study.

My role

I worked with the UX team to define scope and support communication. I focused on craft, gave feedback on design options, and co-designed solutions when the team got stuck. My feedback emphasized information architecture and content over interaction complexity, and making sure activities remained prominent.

Key findings

Top sights & collections together helped users understand a destination and help them narrow experiences.
The components helped users understand our value proposition as a travel companion.
We saw 2 mindsets regarding information — those who wanted to explore activities with additional content & those who wanted in-depth information.

Experiment 01 — initial design

Experiment 1 — city intro and page navigation

The first A/B test kept most of the components from the concept prototypes — a city intro, top sights, and curated collections — with minor adjustments for page navigation and technical feasibility.

We launched in 26 cities with curated collections. The experiment ran for 21 days at the end of 2020.

Unfortunately, the success metrics went the wrong direction — a 10.5% drop in quoter rate (visitors clicking an activity) and 17% increase in bounce rate.

We tested too many competing hypotheses at once. With negative metric impact, it was hard to learn what specifically caused the regression.
Moving activities too far down the page created confusion and mismatched expectations for a lot of visitors who came expecting a bookable marketplace.

Over the next 4 months, the team launched 5 experiments, to isolate hypotheses and learn from experiments.

The biggest challenge was iterating on a highly optimized part of the product while data volumes were suppressed by COVID-19. My role during this time was to help define success and clear hypotheses, and keep motivation up and feedback cycles rapid.

Timeline showing the five iterative experiments across four months

#1 Moving activities up the page

Pushing bookable activities down was causing confusion around our value proposition. We pulled the activity list back toward the top of the page.

#2 Improving the collections

Grouping activities within collections was key to helping people navigate a city. We iterated across multiple cycles — including a "top activities" collection and hand-picking collections for test cities.

#3 Improving the "top sights" module

Travelers anchored themselves through the top sights in a destination. We improved this module and placed it near the top so it was one of the first things users saw.

#4 Interaction & visual design

By the 6th iteration, a mobile-focused usability study drove UI changes to collections and a quick category-filter navigation that made scanning much easier.

Experiment 06 Results

All success metrics in the green.

−3.3%

drop in Bounce Rate — fewer visitors leaving without interacting.

+3.6%

uplift in Quoter Rate — more visitors clicking through to an activity.

+3.14%

uplift in Check Availability Rate — more visitors searching real dates on an activity.

Ways of working

The best approach is a general direction with guiding principles, then testing small hypotheses one at a time.
We only learn from failure if we go into each experiment with clear rationale and intentionality.

Attitude

Perseverance is the multiplier. In an unknown space there will be failure — but when teams have conviction grounded in research, the work feels meaningful enough to push through.

Leadership

As managers, we play different roles depending on what the team needs. Balancing those modes takes trust and swapping tools based on the type of problem.