Optimising the User Experience for EasyJet

At EasyJet, I led the B2C design and execution of a series of A/B tests aimed at enhancing user engagement and conversion rates on the booking platform. My role involved working closely with cross discipline teams to identify key areas for improvement, working with product owners to create a hypothesis, designing test variants, and analysing results to inform design decisions.

Discovery

To kickstart the project, I conducted a thorough analysis of the booking funnel to identify pain points and areas for improvement. This involved detailed documentation of user behaviour, drop-off points, and conversion rates. The outcome of this discovery was that the further along the funnel you go the more likely a user is to abandon their booking. The highest drop off was at seat selection. The assumption surrounding this is that the user is currently researching for the best price. 

Using Userlytics, we conducted discovery research where users were asked to book a flight to a destination of their choice. This allowed us to observe their behaviours and gather qualitative insights into their decision-making processes. To get further qualitative insights I went into Luton airport and asked users to recall how they booked their current flight. The research highlighted key issues and opportunities within the booking flow. It also proved our assumption as I observed users searching for the best possible price when it comes to booking a flight.

I also spoke to members of the cabin crew, the key takeaway was that the main reason flights are delayed is because passengers wanted to sit with their family or travel companion and the last minute shuffling round often causes delays.

easyJet funnel analysis

Brainstorming

With the problems understood I facilitated a brainstorming session. In this session I outlined the problems we were encountering. The workshop gave everyone the opportunity to write down all the possible ideas. I then grouped similar ideas and we dot voted. We then created a hypothesis for the winning ideas.

The winning ideas were:

1. Sit together
2.  Seat selection
3. Fare bundles 

Hypothesis Creation and Prototyping

Together with product owners, we formulated hypotheses aimed at addressing the identified issues. I then translated these hypotheses into ideas and quick prototypes, which we tested at the airport using guerrilla testing methods. This provided us with rapid feedback, which was crucial for iterating on our designs. Additionally, I worked closely with developers to understand technical constraints and ensure feasibility before sharing prototypes with the wider team.

Sit Together Project

Sit Together aimed to improve the seat selection process. 

As a team we agreed on the following:

Discovery and user testing

I conducted a comprehensive study on Userlyticis using unmoderated testing. The aim of the study was to:

  • get insights into users thoughts of sitting together on a journey
  • if the sit together feature is discoverable
  • expected experience on how sit together should function
To get the most out of the unmoderated testing I created a few ideas and iterated over the designs together with the cross functional team. We narrowed it down to one design, but there was a difference in opinion on how the journey should flow through the use case of reselecting a seat. I used the planned session to understand and validate which seat selection behaviour was the expected interaction.
Insights into sitting together
Expected experience

There was a difference in opinion amongst the team, so I wanted to put the prototypes in front of users and get their opinion on the expected behaviour of this feature. A few team members thought when a user selects sits together, if they proceed to select different seats then sit together should be deselected. My thoughts were that if  user has activated sit together, if they reselect their seats then sit together should stay active, unless they deselect it.  As a whole the research supported my design decision, leading to a more intuitive seat selection experience. 

Design

The design below got good feedback from the usability testing sessions and is the one that was developed to be tested. 

 

Results

6.1% increase
In seat attachments
2.6% increase
on progression to the next step, resulting in fewer drop-offs.

Select same seats

Select same seats was voted as another experiment to proceed with. This was to address the issue of seat abandonment and little to no interaction with the feature. This was further supported by my observation that users who selected a seat were missing the opportunity to select the same seats for their return journey. 

 

Hypothesis

“By changing the design and positioning of select same seats and having a clear CTA will increase seat attachments as it will increase the prominence that the same seats are available on the customers return flight.“

Primary KPI:

  • bookings with Seat attachment 

Design

The two variants below were tested.

Variant A – select same seats modal appears on the Outbound page. If select same seats is selected the user goes straight to select cabin bags page. In this scenario the user skips the Inbound page.

Variant B – select same seats will be shown on the Inbound page. If the user clicks yes, then the same seat is selected, they then have to click continue to go to the next step.

Results

4.3% increase
on mobile bookings with seat attachments
3.1% increase
on desktop bookings with seat attachments

Fare Bundles Study

Another area of focus was the fare bundles selection process. Users often struggled with choosing the right fare bundle.  There were also a number of comments left by users during the Userlytics test to say that on mobile it was odd to have the recommended fare bundle in the middle as an option. On desktop the recommended fare bundle is also in the middle, but this is a design pattern that is regularly used. 

To gain deeper insights, I conducted a study to understand holiday makers’ decision making processes, identify the benefits holidaymakers found most appealing, and determine which design was the easiest to digest and most user-friendly.

Prototypes

I did a competitor analysis, comparing how other airlines offer bundles. In most cases the modal was quite different, we wanted to focus on a light touch and incremental changes so that we could understand the impact a certain change had. Before we reordered the fare bundles to show the recommended one at the top or front we wanted to understand what would be the best way of displaying fare bundles. I validated this through Userlytics.

We decided on the following variants:

Variant A – All the packages were displayed vertically. Only the first feature of each package was displayed, users had to click to expand the list to view the remaining features. This was designed so maximise the real estate of the screen, allowing users to easily compare prices and seeing which package has more to offer.

Variant B – Fare bundles were displayed horizontally. All the features of each package were listed out a long with the prices, which meant users had to swipe across to see the other packages.

Fare bundles designs
Fare bundles insights

I created the script and arranged the tasks in Userlytics. After a practice run with the team, I launched the test and analysed the results. The findings include: 

 

Within the study I also did a card sorting exercise. This was to inform the order of benefits that went into each fare bundle. The card sort revealed that the common preferences were:

 

I also tested the interaction for both ideas to ensure users understood they were easy to use and understand. Users understood with variant A, you had to expand the list to see the other benefits, but having the price visible made it easier to compare. With variant B they understood they had to swipe to compare the fare bundle. This made it easier to compare the benefits, but they had to remember the price.  Below are some verbaitum comments from the tests.

Variant A was the preferred design by 80% of the participants in the Userlytics test. Now that we had confirmation that the interactions and design intention was understood by our test group we could launch the A/B test.
A/B testing results

We ended up stopping the test as the control was out performing the Variant. Our assumption was that expanding benefits increased cognitive load for users, therefore having a negative effect on our metrics.

We then moved onto testing the control against one variant. This variant would have the recommended fare bundle at the top of mobile. 

Hypothesis:

Displaying the recommended fare bundle banner at the top of the screen will encourage customers to select this bundle, thereby increasing bundle transactions.

Primary KPI’s:

  • Transactions

Secondary KPI’s

  • Standard bookings
  • Standard plus bookings
  • Essential bookings
  • Revenue
  • Traffic to select seats page

Design

Second fare bundles test

Results

9.3% increase
In transactions
7.2% increase
In standard bookings
4% increase
In standard plus bookings

Design system

To improve workflow efficiency and consistency, I created a design system component library, including page templates. This allowed for the reuse of design elements across various projects, enhancing the speed and quality of our design process. I also initiated a separate project with the developer community within EasyJet to document these components in Storybook. This documentation provided several benefits:

– Enhanced Collaboration: Clear guidelines and documentation facilitated better communication and collaboration between designers and developers.
– Consistency: A unified design language ensured a consistent user experience across all touchpoints.
– Efficiency: Reusable components reduced design and development time, allowing the team to focus on innovation and user needs.

Design system components