Client
Donkey Republic
My role
Strategy
User research
Insights to concept
Prototype testing
UI design
Quality assurance
Year
2020
Challenge
Donkey Republic rents out sharable bikes in European cities to members, tourist and locals. The bikes are picked up and dropped off at hubs in the city. This case focuses on my work with optimizing the bike pick-up flow, i.e. the way users find, select and book bikes. The main goal was to improve KPIs. Specifically, to increase the rate of users entering the booking funnel and lower the cancellation rate. Additionally, to lower the user pains associated with bike pick-up and to increase the NPS score.
Starting point
In the original app design, the user could tab on hub markers on the map to select a bike from a hub. The app-controlled which bike was assigned to the user. Only after completing the booking, the app revealed which bike the user had rented together with details like bike name, distance to bike, and for e-bikes the battery level.
Solution
The redesign of the pick-up flow allows the user to easily select a bike matching his needs, whether standing next to a bike or selecting it from a distance. Exposing information on bike name, distance to bike, battery level, and availability upfront, enables the user to make an informed decision before progressing to the booking. When standing in a hub, the user can choose among the available bikes based on the physical condition of bikes and through the app get information about why a bike is not available. If the user decides to book a bike from a distance, the redesign makes it is possible to plan the trip ahead of time-based on the distance to the selected bike.
Process
The decision to focus on redesigning the pick-up flow resulted from a social mention analysis coupled with considerations on ROI.
I did a social mention analysis to identify issues in the user journey and potentials for business growth. The analysis was based on user feedback data collected from in-app NPS score ratings, app store ratings and social media. These data were analyzed by a artificial intelligence service as this enabled me to find patterns in the large amounts of social data.
The results from the social mention analysis highlighted the pick-up flow as one of the main problem areas. Together with the CEO and CTO, I decided to go ahead with a redesign of this problem after assessing the potential gains on KPIs and the cost of improving essential parts of the pick-up.
The artificial intelligence service also helped identify pains related to the pick-up process, which I could use as a starting point.
To get a deeper understanding of the social mention user feedback, I conducted a series of interviews with different user segments including locals, members and tourists.
The interviews showed that while some of the issues were shared among segments, especially members had unique needs that needed to be addressed in the redesign.
Instead of going straight into high fidelity UI design, I started by creating wire-frames to quickly get a grasp of the pros and cons of different design variations.
The wire-framing ideation was guided by 4 design outcomes, derived from the social mention analysis and the 1-on-1 interviews:
- User-determined bike selection
- Easy selection when standing next to a bike
- Visibility of bike details and availability before booking
- Visibility of bike position
The wire-framing ideation resulted in a series of flows exploring possible map navigation and map representations of bikes and hubs. I also considered different ways of using Bluetooth or QR scanning to enable easy selection of bikes standing next to you. While having great promise for the user interaction, we had to discard both scanning solutions due to the lack of Bluetooth detection precision and the high cost of mounting QR codes on all bikes.
The ideation process ended up in two high fidelity UI flows, after elaboration with the team.
In many projects, I would have opted for a quick test on a low-cost bare-bones prototype. However, in this case, I decided to go straight into development to test the two high fidelity flows. The main reason for this decision was a realization that the only way to gauge the feasibility of the solution would be through observation of how users behave in a real-life setting booking actual bikes. Of course, this was risky because the prototype tests might send me back to the drawing table. However, we already had a great part of the technical architecture in place, which meant that building the two prototypes would only take 2-3 days.
In total, I ran approximately 10 tests on the two prototypes with different user segments. The prototype tests indicated that one of the prototypes outperformed the other one across user segments and highlighted a couple of things to change in the final implementation.
Outcomes
Before launch, I set up KPI dashboards to monitor the effect of the update together with our BI.
Out of the total users opening the app, the post-update data showed a 5 percentage points increase in users entering the booking funnel, compared to the same time last year. The cancellation rate dropped by 6 percentage points while time spent from finishing the booking funnel until unlocking was lowered by almost 2 minutes. In total, the improvements on both pre- and post-pick-up reflected in the NPS score by an increase of 6 percentage points.
