The case for changing:
addressing cancellation
costs

How do you make changing a booking as easy as cancelling one? I worked across the cancellation experience, booking rules and longer-term servicing strategy to help travelers adjust their plans without losing their trip.

Role

Senior Product Designer

Team

Content, PM, Engineering

Cancellations−16.68%
Projected annual figure. Test traffic compared with all other bookings.+$1.12M
Support time saved$174k
Expedia change-your-stay flow: change dates, pick new dates, review, and a 'Your dates have been changed' confirmation

Future state: change dates flow, by Ashley Riehlin.

// Where we started

Single-click cancel

This project started from an earlier win.

Cancelling used to be a common frustration, so Expedia made it a single click. Travelers loved it: more of them sorted things out themselves, calls to customer support dropped, and more than 1 in 10 re-booked the same destination within a week (59% of them the same hotel).

But cancellations went up 1%. Could we steer travelers towards changing their booking instead?

Manage booking screens for a stay in Chicago, with a Manage booking button
// Spotting patterns

Why do travelers cancel?

I worked on a taxonomy of cancellation reasons, drawing on Voice of the Customer data to understand what was driving travelers to cancel their trips.

One thing stood out: many didn’t want to cancel at all, but make a change instead. Add another guest, or move their beach holiday to the following weekend when the forecast looked better.

That got me thinking: what would it take to let more travelers change their bookings directly in the app?

A full redesign of the experience and underlying architecture wasn’t in scope (yet). So where did I start?

My requirements have changed

My nice has decided to join us on our trip, I need to adjust my booking to accommodate her and her needs. It is unclear to me if I can make the changes I require or not.

Sort / hard change + flights

Reason for cancel (FE)

L1L2L3

I need a different solution

Room

Room type
Bed type [..+ others]

Property

I need a different location
Does not cater to my needs [..+ others]

I am not convinced by this property anymore

I have done more research since booking this property, and the reviews do not convince me. The host / property is not responsive and I do not trust this property.

Reason for cancel (FE)

L1L2

I am unhappy with this property

Bad reviews
Low quality
Property / area not safe
Property is non-responsive

Taxonomy excerpt: sentiment / scenario.

// The problem, reframed

At the time, changing a booking took more effort than cancelling it

Cancelling took a single click.Changing meant hunting through the booking details.
Technical investigations found out 36% of bookings couldn’t be changed online at all.Resulting in support contacts.
// What we tried

How might we make change options visible?

Cancelling was one tap; change options meant hunting through the booking details. Redesigning that specific flow was left out of scope, as it was deemed outside our team’s responsibility (..I have a lot of thoughts on this).

Instead, I looked at the areas we could realistically have an impact on and move quickly with, the first one being the cancel flow itself.

Hand-drawn cancel flow sketch overlaid with a decision diagram: is the booking eligible for change, intercept the cancel, inform about change options, or re-shop
// Testing approaches

How much friction is too much friction?

TL;DR: travelers weren’t fans of being asked why they wanted to cancel, before the fact. Even less so if at that point they weren’t presented with another option. Some were concerned their answer could have negative repercussions on their future bookings, others simply didn’t want to justify themselves, and rightly so.

Ask why they’re cancelling and surface the relevant change option.

The idea behind this is that it would help us collect more data. A lot of doubts followed.

Prototype: cancel sheet asking why you’re cancelling, with a banner saying the booking is eligible to change without fees

Intercept the cancel, and inform their booking was eligible for a change.

Lighter, less decision making upfront, but less data collection on our side.

Prototype: cancel reservation dialog offering to change dates instead
// What we decided

Evaluation

Asking “why” on its own was the least-liked option in testing, while the nudge to change landed well.

At the time, the change capabilities weren’t fully built, so we couldn’t be specific about what the change options were, as that logic lived in another part of the flow.

Asking why without offering another option wouldn’t reduce cancellations (our primary metric) and could hurt repeat bookings, so I made the case for not building it on its own.

The order in which we decided to build “the thing” was as important as deciding what not to build.

Would we want to build this…
Prototype: reason sheet with a single list of reasons
…if it didn’t have this?
Prototype: the same reason sheet with change options offered under ‘I cannot travel’

What would make changing a booking possible, before making it easier?

And how could we start making progress without a complete, end-to-end redesign?

// Delivery

Quick wins and the long game

Unlocking non-changeable bookings

Enable an additional 36% of travelers to self-serve lodging changes by removing legacy booking restrictions and introducing a flexible front end that would dynamically communicate the impact of each change (on payment methods, coupon use, cancellation policies etc.).

Results
  • $174k support time saved
  • −16.68% cancellations, equating to an estimated $1.12M a year in revenue kept*
What to know before changing your dates: heads-up screen for previously ineligible bookings, mapped against booking edge cases

Nudge to change

The “would we build this if we couldn’t build that” argument won. We built a gentle prompt to change over our cancellation dialog, so travelers see the options that fit their situation at the moment they’re about to leave.

Projected results
  • −16.68% cancellations
  • About 9% of travelers who start to cancel tap “Change my booking”, and about 58% of those finish a change
  • About 95% still cancel, with no rise in contacts to support
Cancel reservation

If you cancel now, we will issue a full refund of $987.51 to your original payment method.

Are you sure you want to cancel your reservation?

Cancel now
Change your dates instead?You can move around your plans, without fees.Change my booking

Reason for cancel survey

Voice of the customer was good for an initial read on the issue and to start brainstorming ideas, but we needed quantifiable data to make our case for change stronger. Based on the findings from the approaches testing, we launched a post-cancellation survey to help us with that.

Before you go, tell us why you cancelled. It takes a minute.

How satisfied were you with your stay?

1234567

Why did you cancel your booking?

Anything else you’d like to tell us?

Submit

A help centre for every booking

I left the team with a vision for one place to manage every booking, adapting help, changes and cancellation options to the traveler’s needs from booking to check-in.

Handed over

Delivery and discovery boards: small tactical wins with test results, and benchmarking plus a service and system map for the servicing hub
// Learnings

What I’m taking with me

  1. 01

    Make the better path just as easy

    Removing friction from cancelling worked, a little too well. The answer wasn’t adding it back, but making changing as easy as cancelling.

  2. 02

    Make a question useful

    Asking “why” on its own felt like a hurdle. Adding it alongside a real alternative made it feel like help.

  3. 03

    Unblock before you build

    Removing old rules that stopped bookings from being changed beat any new feature, and gave us the evidence to make the case for a single help centre.