Since early 2015, the Egencia Homepage team was tasked with migrating the codebase from multiple legacy platforms onto a single, global platform. This massive effort took over three years to complete and required the full attention of the entire team. This commitment meant that there was little time or energy left over to devote to developing a product strategy, which led to a disjointed experience. It became clear that this lack of vision meant we weren't doing a good job of helping users with task completion or efficiency.
As the migration project came to a close, we began to feel a combination of excitement and anxiety about what lay ahead for our product. What we needed was a new vision, a strategy to take us beyond migration and into a strong future.
In August of 2018, I facilitated a Design Thinking session involving half a dozen different functions from across the company. Alongside Homepage product managers and engineers, we had members from Account Management, Sales, and Leadership, as well as designers and product managers from multiple other products. We felt strongly that in order to develop a well-rounded story, we needed to include as many perspectives as possible.
To practice building empathy, the session started with partner interviews. Since everyone in the room had traveled for business previously, it was an easy topic to talk about. Subsequent exercises were aimed at building upon these stories and gathering insights. Soons, fairly striking patterns began to emerge. Here are some examples:
Business travelers are inherently different than leisure travelers. Leisure travelers are motivated by price, whereas business travelers have a variety of motivating factors, including personal travel points and status, office proximity, and company policy. These differences became more apparent as we continued the conversations throughout the session.
These HMW questions guided each group in another round of ideation. Hundreds of ideas were generated, creating a truly wonderful and colorful sticky-note graveyard. The best part about the ideation exercises was that the group really stretched outside of the box, coming up with things that normally wouldn't see the light of day.
Our final exercise, as photographed above, helped put a framework around the spectrum of ideas. From the lens of implementation, the first category was Incremental, the next, Evolutionary, and lastly, Revolutionary. Each group worked together to categorize their sticky notes accordingly. These conversations helped everyone understand technical limitations in a way that grounded their ideas in reality.
One we needed to remember was that many of these ideas could not be controlled by our team. However, Homepage is a user’s first touchpoint into the product – for all users – so the goal for the session was to see which aspects were important for business travelers, and how Homepage can facilitate in making those happen. The brainstorming session proved to be invaluable in our quest to define a product strategy. We spent a handful of days afterward synthesizing all of the themes and ideas that came out of the workshop.
We also collected data to supplement all of our new ideas. The metrics below are a couple of the primary examples. What this data tells us is that our users vary from true power users to people who only visit the site a few times per year. We needed to create a product that addresses the needs of this broad spectrum of users without negatively impacting their experience.
With all of that data combined and considered, we distilled everything down to our primary focus. And we wanted to answer this question by doing 2 things.
Over the years we've collected anecdotal data that suggests our user types aren't all coming to the website with the same tasks. There's actually quite a range of actions a user needs to do. With that in mind, we realized that the homepage does not facilitate the most efficient way to accomplish those actions. Our user types have vastly different goals, but our product has never considered those differences.
Up until now, the Homepage had kinda functioned as a dumping ground everything, but without any real thought or direction behind what got put on the page. So the new direction is not only a new design but a massive data science effort. We teamed up with the data science team to utilize machine learning that will help us to leverage users' past behaviors to help them complete future tasks. We're hoping to empower our users to complete their day-to-day tasks much faster with less friction.
Wireframes shown below illustrate various tasked surfaced to the top of the page depending on the user type.
Test 1: The first AB Test we built drove us straight into our vision. We introduced the Hero section, queuing our users into changes ahead. All content surfaced into the dark blue area will be dynamic, relevant, and actionable, but it will not exist unless there is something to do. If there is no actionable content, the default would be the Search form. This is an exciting moment for homepage because we are starting to contextualize the product and make it a little more user-friendly.
Another goal of this design is to help users more easily access their profile page, and in the future, we can start playing around with things like document and credit card expiration warnings, nudges to finish filling out profile information, and more.
Test 2: Immediately following the Hero test, we tested the first round of actionable content – Upcoming Trips. Since we know that most of our users only travel 1-3 times per year, as a user with a trip in the near future, it's more likely that a user will want to interact with that itinerary than start booking a new trip. Plus, our Road Warriors (people who travel 12+ times per year), will have better insight into their trip cadence. This module was tested above the search form inside of the Hero area. In the images above, you can see the control for this test, with the Upcoming Trips module below the search form.
Test 3: One of our customers' biggest points of frustration is related to receipts. From a back-end perspective, the payment systems are incredibly complicated, involving multiple players and merchants of record. When it comes to expensing a trip though, users don't care how complicated the payments process it is – they just want to be able to find their receipts in order to get their money reimbursed.
After the Upcoming Trips test (above) was rolled out as a new feature, one of the enhancements we decided to test as a follow-up was around addressing this problem with receipts. The idea is that we would maintain a past trip in the Upcoming Trips section with a direct ingress into their receipts.
Below are the current (March 2019) solutions for the first big push toward our new direction. The product culture at Egencia is incredibly iterative, which means we'll be moving quickly toward this goal, and moving beyond it within the next 6-12 months.
This new strategic direction has helped bring purpose to the product. We’ve been able to focus all the work and feedback on whether or not it aligns with our goals. We’ve really only started with a small subset of features and ideas, and there are so many more out there. When we look at the user journey, it can be bucketed into three main areas: pre-trip, during trip, and post-trip. But when we cross that with each user role type we can immediately see there's tons of opportunity to continue personalization, making each user’s homepage their own.
All of the tests we've launched have been feeding a model we're creating with the data science team. In a few months, we want to be able to state with data which actions are most important to our users with a specific order of importance. Then use this model to influence page layout at a user level.
Despite the research and anecdotal data supporting our hypothesis, the AB testing data seems to suggest that we were incorrect to move the search form lower on the page. Even though users can have many tasks when landing on the Homepage, the most important task still seems to be searching. We are still investigating the root cause of the data trends we are seeing, but it looks like any content we test in the Hero area causes a dip in conversion. Going forward I plan to lean more heavily into user research by working with our actual users (not just usertesting.com). Given that we were all a little surprised by the results of the first round of tests, this would help better prepare us in the future.
On the bright side, I helped deliver some big wins thus far: