One of the top priorities for Uber in APAC was improving the pickup experience for riders and drivers. In complex and unfamiliar environments, being able to get a ride reliably was paramount to delivering a stress-free transportation experience. No matter if you were in a new city for the first time, or trying to get a car for your morning commute, we wanted you to be able to simply “push a button, get a car.”

Uber’s original pickup experience relied on a rider moving a pin on a map: while this worked great in the US, the experience broke down in Asia where pickups often do not happen on the street — but rather in designated pickup areas embedded in complex, multi-storey urban environments. Simply providing a pin and street address didn’t give drivers enough context to meet their riders — an experience that often resulted in stressful phone calls and missed pickups.

“Venues”

In June 2016, I visited Singapore to help prep the Uber Southeast Asia team for the upcoming launches of uberPOOL in Singapore and Manilla. I was staying at a hotel in the Orchard district in Singapore — a major shopping and commercial district. My hotel’s street address was on Orchard Road; however, the road was a fast-moving, busy thoroughfare where the sidewalk was fenced off for pedestrians, and stopping on the side of the street was illegal for cars.

In fact, the hotel had a designated pickup & dropoff area in front of the lobby; but drivers could only access it from a different street — from a perpendicular road (Scotts Road). And if a driver missed the entrance, he would have to drive around a loop that would add 5-6 minutes to the pickup — costing the driver additional unpaid time, and adding stress for the rider who now has to wait twice as long for his ride.

Uber’s original solution to address complex pickup environments was the addition of “pickup notes” — allowing riders to type a short note to the driver (i.e., “I’m in the hotel lobby; access from Scotts Road”). However, this was far from an ideal solution — riders would have to type a note everytime they requested a ride, and none of this information was passed in to the driver’s navigation. The amount of international travellers relying on Uber made this problem worse as drivers often couldn’t understand the language that their riders were communicating with them in.

When I left Singapore to return to China, I wrote a note to the Singapore team about this experience and suggested a different solution: adapting the tooling Uber had built for airport pickups to urban environments.

Major urban complexes in Asia actually had many similarities to airports: for starters, there were designated pickup points that riders and drivers were obliged to meet at. Further, riders and drivers were both often perplexed as to where they should meet — a rider going to an unfamiliar shopping mall for the first time may not know where to request their ride, and a driver that hasn’t picked up anyone at the complex may find it challenging to navigate to the pickup location.

Uber had built a customised pickup experience for airports specifically, allowing global operations teams to designated pickup zones (i.e., terminals) and pickup points (i.e., door numbers). While this was live globally, it was only used at airports.

Later that year, the Singapore operations team started deploying this feature — internally called “Venues” — across a number of major shopping malls and office complexes.

When the Uber app detected riders were requesting from inside a Venue, the rider would be asked to select a pickup location, which would be passed to the driver. Riders were guided with wayfinding information, and because the pickup location was controlled entirely by Uber, we could ensure that critical information (i.e., “hotel lobby” or “level 2 carpark”) would be correctly translated and integrated into the driver navigation flow.

A quick comparison of marketplace metrics before and after this change revealed double-digit improvements in pickup times and trip completion rate.

“Hacks” and Hotspots

Of course, this solution was not perfect; at Uber, we called these informal product adaptations “hacks” — and for good reason. Venues at times could add friction to the rider request flow, and it was difficult for a rider who may have been erroneously placed into a Venue — for instance, if their GPS location was wrong — to escape out of the flow.

The majority of the efforts in San Francisco still revolved around improving the pickup experience in the US, and continued innovation focused on the curb-side pickup experience that riders in America experienced every day. This ultimately meant that the Asian product experience for Uber riders and drivers was “branched off” from the efforts of the global team.

In early 2017, Uber launched a new rider app (internally codenamed “Helix”). One of the core promises of this new rider app was that riders would not have to manually set their pickup location — Helix was “destination-first” in that riders would only have to tell the app where they were going, and the app would intelligently determine where the rider was and suggest a pickup location.

To help deliver on this user promise, a considerable amount of effort was dedicated to generating machine-learned “Hotspots” — pickup points based on where riders and drivers historically met each other.

The problem was that these Hotspots were generated without contextual labels, and only referenced as latitude & longitude coordinate pairs displayed as a dot on the map for riders, or reverse-geocoded street addresses for drivers. The Venues that we created were entirely incompatible with Hotspots — meaning that the advanced machine-learning algorithms would not benefit the pickup experiences we needed them to help with the most.

Don’t ship your org chart

By mid-2017, I had asked each of my country product leads in Asia to implement Venues across their markets — in markets like Vietnam and Thailand, close to a thousand were created within a matter of days.

However, this clearly wasn’t sustainable; the right answer for Uber was to develop a solution that integrated the contextual information provided by Venues with the machine-learning provided by Hotspots — and eliminate the incompatibility between the two features.

Unfortunately, Venues were owned by a team whose charter was to improve the pickup experience at Airports — and not the shopping malls, office complexes, or large housing estates that dotted Asian cities. The Asia team’s pleas to the Rider team, who largely owned the Pickup Experience, were mired in cross-functional communication frictions and progress was slow until San Francisco proposed to create “pickup zones” for Uber in certain neighbourhoods. A cross-functional effort was kicked off to build a better solution than the existing Venues experience provided, incorporating some of the initial research that my team conducted in Asia a year prior.

The challenge throughout this year-long ordeal for Uber in Asia revolved around the regional focus on our customers, rather than our product and engineering team’s org chart — we were, as a regional team, eyeballs-deep in the needs and wants of our riders and drivers, but often unaware of the complexities of navigating the 100+ tech teams situated in San Francisco — and the challenges of collaborating across teams (who often were re-orged on a quarterly basis).

Ultimately, a quote from a senior leader in our product org resonated with me: “don’t ship your org chart.” As a product leader, your responsibility is to build on behalf of your customers — capturing their needs and wants, and distilling that into innovative and delightful experiences. All of this is irrespective of the artificial boundaries that your org chart puts on your team; reaching out across functions and collaborating — and even toe-stepping (another Uber cultural value) — is critical to delivering the best experience for your users, especially in larger companies where team structure and feature ownership becomes more complex.

Uber’s business in Southeast Asia was sold before a longer-term solution could be built. However, Grab has taken the reins of improving their pickup experience in mirror to our efforts from 2016; I’m excited to see where they take the urban pickup experience.