Platform: iOS and Android
Scoop is a carpooling app that enables colleagues and neighbors to easily arrange carpools to and from work. Unlike on-demand ride-hailing apps, Scoop users schedule carpools in advance. The app matches them with another person traveling a similar route at the same time.
We set out to upgrade the carpool scheduling experience to give users more control over their pickup time and reward them for being more flexible. This resulted in a 47% increase in time flexibility for new users.
When Scoop users schedule a carpool, they have the opportunity to set a pickup time window, which is the interval of time in which they are willing to pick someone up or be picked up for their trip. A long pickup window is great for the carpool community as a whole, because the more time windows overlap, the more carpools can successfully be arranged. For carpoolers, entering a long pickup window has a significant downside: they have less control over their schedule.
Our high level goals were to:
1. Improve the carpool scheduling experience by allowing users to specify both a pickup time window they could commit to as well as a single preferred pickup time, and make their carpool as close to that preferred time as possible.
2. Improve Scoop's ability to match more carpools by encouraging users to commit to longer pickup windows.
Illustration by Siyang Liu
I was the sole designer for this feature. I worked closely with a product manager to define user needs and I designed the app experience. I reviewed past feedback from users, created sketches, prototypes, flows, and pixel perfect designs. I collaborated with Android and iOS engineers on implementation and QA. I documented new components and added them to our design system. For carpooler insights gathering and live user testing I worked alongside a UX researcher. I partnered with a data scientist and a product manger to evaluate the feature's performance.
After reviewing our customer support tickets and talking to Scoop employees in data science, support, UX research, and product management we identified these themes:
1. Carpoolers wanted more control over their carpool pickup time. Up to 40% of 1st time rider cancellations were due to the rider not getting the carpool pickup time they wanted. And we had a high amount of first time users who were canceling.
2. Carpoolers thought it was more important to get matched in a carpool than to have their carpool at the exact time they wanted. The average length of pickup time windows chosen by carpoolers was 32 minutes. Increasing this time would allow Scoop to match more users in carpools.
3. Carpoolers who might be willing to be flexible once in a while ended up shortening their pickup windows over time because they were consistently getting assigned carpools at a time that was not their preference.
Based on these goals, user needs, and business needs, I explored several concepts in sketches. I start sketching by writing out the conversation I want the product to have with users, and then I turn each step of the conversation into an idea for a piece of the UI. The pieces are arranged in the order the conversation would naturally happen, with progressive disclosure to focus the user on the task at hand. After sketching out some concepts I arranged them into flows to share with my team.
This is one of the flows I created out of early sketches:
After gathering feedback from the team I developed two versions of the design into high-fidelity prototypes for usability testing. We recruited current users of Scoop as well as folks who had not used the product to go through two prototypes live with a researcher. All participants saw both prototypes and we alternated which prototype was shown first.
In this prototype, users start by setting their pickup time window, and then use a slider to indicate a preference for "on the earlier side," "near the middle," or "on the later side."
Pros:
In testing, users understood the benefit of being able to widen their pickup window to increase their chances of getting matched in a carpool, while also being able to communicate their preferred time.
"This is a great feature, because I often get matched on the earlier or later sides of my range, and sometimes I don’t want that, but it’s better than not getting matched. So, this is kind of cool.”
Having only three preference options to choose from (earlier, middle, later) felt cognitively easier than a selector that allowed for more specificity.
Cons:
Specifying a time range within the original window wasn’t quite getting to the crux of the matter: control.
More pros:
When users had selected less than a 45 minute time window, we disabled the time preference by reducing the selector opacity and putting guidance on the screen about how they could enable the feature.
The visual representation of what would be available helped users to understand what they were missing out on and that they could enable the feature by taking a specific action.
I organized the prototype screens in the same order that they would be in InVision prototyping software, with screen naming designed to make it easier to add InVision hotspots.
In this prototype carpoolers started by setting their pickup time window, and then used a slider to indicate a specific preferred time.
Pros:
Users felt this increased control over their schedule without any negative influence on matching expectations. Even with the specific time preference, they understood that they had committed to a pickup anywhere in the time range.
Some users increased their time range after seeing what the feature did.
Cons:
Some users felt that this was a more complex decision that required more thinking than the earlier/middle/later way of selecting in the first prototype.
In this version of the state where the user had less than a 45 minute pickup window selected and therefore could not use the time preference, there was confusion. Users understood the message to mean that they needed at minimum a 45 minute time window to schedule a carpool at all.
To combine the strengths of both versions, we made these design changes:
We kept the specific time selection, but reduced the number of options available by providing 15 minute increments.
We arranged the time preferences on a horizontal scroll, to make sure that very long time windows with many 15 minute increments didn't push the blue selection button off the screen.
We took great care in implementation to manage the discoverability of the horizontal scroll. We made sure that for all common iOS and Android screen sizes the first time preference button that went off the right side of the screen was partly visible.
One tap on a 15 minute time interval selects the time preference.
When the pickup window is less than 45 minutes, we kept the time preference selector visible but disabled.
In an earlier round of user testing this feature we compared different versions of the time picker mechanism and focused on ease of use.
After seeing the test results, we went with a time picker that most closely aligns with the standard iOS pattern for time pickers and used it for both Android and iOS.
While we knew that we were making a choice to not use native Android patterns for the date picking, we didn't realize until the implementation phase that we would lose the built-in accessibility features that come with native Android time picking interactions. Our Android engineer had to build the screen reader capabilities from scratch.
After the time preference feature went out to Scoop's 18,000 weekly active carpoolers, we had one piece of feedback that we heard several times in interactions that carpoolers had with our Support team.
While choosing a time window, users didn't find it intuitive to close the second time picker after making their selection, which left both the time preference and the "Choose" button off the screen on smaller devices.
We couldn't collapse the time picker by default because that would assume users only needed one tap on the wheel picker to get to the right time.
The back button wasn't designed to save changes. This was a problem because in this state, it is the only visible way out of the screen.
We tried pinning the Choose button to the bottom of the screen on a fixed footer, but this negatively impacted the discoverability of the time preference setting, as it was off the screen on small devices. We saw a dip in the percentage of users utilizing the time preference feature.
We reverted this change and updated the setting on the back button to save users' edits to their pickup time.
Saving changes automatically with the back button worked well because users still had another chance to confirm their pickup time before scheduling their carpool on the previous screen, the scheduling summary screen.
Three months after the launch of the time preferences feature we saw a 21% increase in carpooler time windows, with larger and more immediate effects on new users of the platform. New users had, on average, a 47 minute time window, up from 32 minutes, a 47% increase. At this time Scoop's user base was growing very quickly, and a large effect in new users was a huge win for us. As time went on we also saw continued gradual increases in pickup window flexibility for existing users.
The success in this project led us to invest in future projects to give users even more visibility into how they could increase their likelihood of getting matched in a carpool.
We received great feedback about the extra control that carpoolers felt they had over their carpool time, and our matching algorithm was able to create more ideal carpools for users.
"If I put a small window [before this feature was available], it was probable that I wouldn't get matched. So I set my window to the maximum in hopes of getting any match. This helps me to make sure I’m still able to prioritize the time I actually want.”
- Carpooler
