Lyft Rider Redesign

Design Goals

Fast + Simple

• Optimized for price/time comparison

• Core actions are reachable and requesting a ride requires fewer taps

CONTEXTUAL

• Surfaces the right recommendations for different times and places

• Repeat rides are a breeze

WORKS FOR EVERYONE

• Legible, accessible, and scales across devices and platforms

• Accommodates future modes of transportation

Product goals

• Ensure metrics parity between the control treatment and the redesign before rolling out to 100% of users on both platforms.

• Continue to iterate and improve existing features. Validate UX changes with data.

• Continue to work towards feature parity — feature parity is not a blocker for launch.

In the previous app design, you had to enter pickup and drop-off in 2 separate steps. There was no way to add a third stop before selecting a mode. You also could not re-order your destinations.

OLD: START TO FINISH

Previously, riders were asked to set their pickup location before selecting a destination. The default view was a search field, rather than a map view, so they were unable to see their GPS indicator and verify that their GPS location matched where they wanted to be picked up.

NeW: DESTINATION FIRST

Now, riders set their destination first, have the ability to add a stop, and confirm their pickup location on the map. Power riders can also bypass manually entering their destination by tapping a shortcut on the Home screen.

MY WORK:

Search UX/UI improvements, panel transitions, search field states and transitions, search recommendation logic improvements

MY TEAMMATES’ WORK:

Destination first flow, personalized home greetings, editable shortcuts, branded search results

UX deep dive

1. Adding a stop before requesting a ride

2. Adding a stop in ride

3. Removing a stop

4. Editing Pickup before requesting a ride

5. Error - no search result selected

In the previous app design, you had to choose a mode first before entering your destination, and there was no easy way to compare price and time across modes. Riders would have to tap on their previously selected mode to pull up a second variation of the mode selector, updated to reflect price based on destination.

OLD: START TO FINISH

Previously, riders were asked to choose what type of car they preferred and set their pickup before choosing a destination. While this flow mapped to the mental model of inputting trip directions, Lyft was unable to provide riders with price and time information that would allow them to make the right choice for their immediate needs.

NeW: DESTINATION FIRST

Asking riders to set their destination first allows Lyft to expose real-time price estimates and ETAs, allowing easier comparison between transportation modes.

MY WORK:

Interaction pattern audit, mode selection and map transitions, UI polish

MY TEAMMATES’ WORK:

Multivariate experiments for mode content layout, integrating new modes, integrating scheduled rides and editing payment flows

MY WORK:

Drag pin on map interaction, panel & map transitions, UI polish

MY TEAMMATES’ WORK:

Pin visual and interaction refinement, walking indicators, far pickup detection, pickup notes

The experience beyond selecting a mode and requesting a ride is what we referred to as the post-request experience. This became my primary area of focus as we worked tirelessly to achieve metrics parity between the control and the redesign. I designed a number of experiment variants for several features before landing on the versions that shipped.

MY WORK:

Panel framework, panel and map transitions for Classic Lyft, multivariate experiments to reduce cancels

MY TEAMMATES’ WORK:

Matching states for different modes, mode illustrations

The In-Ride panel has 2 states, collapsed and expanded. The collapsed state displays only the information that is essential for a successful pickup. The expanded panel is where all other ride-related content lives. This distinction is important for preserving functional map real estate, so that riders are able to interact with the map as they are figuring out where to meet their driver.

MY WORK:

Panel framework, panel & map transitions, driver card

MY TEAMMATES’ WORK:

Trip & payment cards in the expanded panel

Panel guidelines

To support the needs of our rapidly expanding design team, I created design guidelines for an extensible panel framework, to enforce information hierarchy at critical moments in the post-request flow and preserve map functionality.

Post-request cancels are the most costly for both drivers and Lyft as the driver has invested time, gas and mileage heading to a pickup spot before the rider cancels. Riders are prompted to answer a short post-cancelation survey asking why they canceled the ride. From these survey results, we saw that a statistically significant number of riders were canceling because they were unfamiliar with how to edit their trip details once the request had been made.

In an effort to address these comprehension issues and dissuade riders from canceling after a driver is already on their way, I designed a series of experiments, including this (winning) variant which bundles key ride actions into a bottom sheet.

MY WORK:

Edit ride action sheet, cancel confirmation variants, edit pickup/drop-off/add stop flows

MY TEAMMATES’ WORK:

Send ETA flow, actions for non-Classic Lyft rides, additional cancel friction experiments

Drivers who meet certain requirements are given a dashboard-mounted light-up device, called Amp, which when paired with a driver’s phone will display a color that matches the color riders will see represented in their app. This can be especially useful when riders are waiting to be picked up at a crowded venue and have trouble reading license plates at night.

My goal was to translate this feature into our new design language and increase riders’ comprehension of Amp.

One of our learnings from user research and experimentation was that riders tend to disassociate the act of tipping from how they felt about their overall experience. So even riders who may have not had the most ideal experience (ex: a lot of traffic, drop-off time increased, etc.) were still likely to tip their driver. Due to this finding, another team devoted a significant effort to improving the ratings experience, while our team focused on the tipping experience.

Our team’s goal for the end of 2018 was to increase driver tips by 20% — a goal which we actually exceeded by 1.5%, through a number of iterative improvements. In addition to several small behavioral nudges, I designed a feature that allows riders to add a tip during their ride.

MY WORK:

In-ride tipping flow, post-ride tipping experiments

MY TEAMMATES’ WORK:

In-ride ratings, post-ride tipping & rating experiments