INTRODUCTION

Imagine if Uber, Doordash, Amazon, & Instacart is combined into one super-app. That’s Gojek.

As a Product Design Intern within Gojek’s Transport Team which manages the ride-hailing services for cars and motorcycles, I had the rare opportunity to address a critical issue: high cancellation rates during the ride-hailing process. Under the guidance of my mentor, Idwan Ula, I led the design efforts to tackle this issue.

This project was part of an ongoing initiative to improve the fundamental quality of the live tracking experience and was a key focus for Gojek in Q3 2021.

PROBLEM

High cancellation rates during the ride-hailing process were a significant concern for Gojek.

Each ride cancellation not only disrupted the driver’s schedule but also negatively impacted the user experience, leading to frustration and decreased trust in the service. Additionally, high cancellation rates resulted in millions of lost revenue and reduced overall efficiency for the platform. Hence, we initiated this project with the aim to user anxiety and cancellations during the pickup phase.

OBJECTIVES

Our goal was to reduce cancellations during the driver's journey to pick up by achieving these two goals:

Reducing user anxiety

Building trust and reliability in Gojek's services

RESEARCH

We set out to understand when and why users cancel their rides.

To uncover the root causes of high cancellation rates, we employed a variety of methods:

• Surveys & Interviews: Gathered quantitative and qualitative data from the survey in the cancellation flow as well as virtual userbase interviews to understand their experiences, expectations, and reasons for canceling rides.
‍
• Journey Mapping: Visualized the entire user experience from booking to ride completion to identify possible pain points and moments of frustration.
‍
• A/B Testing: Compared different versions of the live tracking screen to determine which elements improved user satisfaction and reduced cancellations (Indonesia only).
‍
• Contextual Inquiries: Observed users in real-world settings to gain in-depth insights into their behaviors and challenges during the ride-hailing process (Indonesia only).

WHAT WE FOUND

Users are most likely to cancel orders during high-stake trips due to various anxiety triggers.

High-Stake Trips: Users in Indonesia (ID) and Singapore (SG) often experience pickup anxiety during high-stake trips, such as commuting to work, rushing to appointments, or navigating first mile/last mile (FM/LM) connections, especially during busy weekday hours. These trips involve significant responsibilities and time sensitivity, making delays particularly stressful.

Anxiety Triggers: Participants reported several anxiety-triggering situations during the ride-hailing process such as:

High ETA

Expected at busy hours, but still causes anxiety.

Driver not moving

When the driver's location seems static.

Driver distance too far

Often when ordering from residential areas.

Long actual waiting time

When the actual wait time is longer than the ETA.

KEY INSIGHT

3 key reasons behind most cancellations.

Lack of real-time information

Users are unaware of real-time changes, causing concerns, especially when drivers appear stationary.

ETA & distance disconnect

Users are more likely to cancel their rides even when drivers are close in proximity if the ETA is still high due to traffic conditions.

Perceived long wait times

High, precise ETAs led users to believe they would wait longer than necessary, increasing cancellations.

APPROACH

Finding the right balance of effective, simple, and cost-efficient.

We explored multiple design approaches for this project such as preset driver messages, where drivers use preset messages to keep users updated through the chat feature, and a robust push notification system to keep users updated. The preset driver messages, while personalized, added to the driver’s workload, which can be dangerous on the road. The robust notification system, although informative, risked becoming intrusive and overwhelming.

After evaluating various options, we decided to focus on microinteraction as it offered the best balance of simplicity, contextual relevance, and cost-efficiency while still allowing us to address user pain points in a manner that was both efficient and impactful.

TRIGGERS & RULES

Defining the conditions & framework for effective microinteractions.

Triggers are used to address specific situations that users commonly encounter during the ride-hailing process, while rules help manage the complexity and relevance of the information presented to users. This end logic was achieved through collaboration between designers, researchers, data scientists, and backend engineers.

Driver's distance

When the ETA is high but the driver is close in proximity (≤500m or ¼ of original pickup distance).

Driver's location

When the driver's location is stationary for an extended period of time (≥60s) or it cannot be fetched.

Deviation from expected path

When there are any deviations from the expected path on the map.

Traffic conditions

When traffic is causing the driver’s location to be stationary and/or increase the ETA by ≥120s.

High stake trip identification

Elements will be shown more frequently when users are on high-stake trips (e.g. commuting to work). This is based on predictive data derived from personalized user patterns, order history, and saved addresses.

Consistency and predictability

Elements must behave consistently across the app. For example, the same type of notification or update should always appear in the same manner, so users know what to expect.

FINAL DESIGN

Four design solutions.

Dynamic status pill

Appearing and disappearing based on real-time triggers. This pill uses the predefined triggers to display relevant information, such as driver's distance, delays, or route changes. A subtle animation and a consistent location ensures that the pill catches the user's attention without being intrusive.

Traffic light tooltip

Appears on the map beside the vehicle icon, providing real-time updates about traffic light stops. This tooltip is triggered when the driver is stationary at a known traffic light location, based on the Indonesian database of traffic lights. This tooltip helps users understand why the driver is stationary. This feature exists separately from the dynamic information pill to avoid redundancy, given the frequency of traffic lights.

Distance Progress Bar

Displayed on trip summary and when users are about to cancel. This feature includes a progress bar with a moving vehicle icon and real-time distance information. The distance progress bar reassures users of the driver's proximity and helps reduce cancellations.

Display ETA in range

Shows the estimated time of arrival as a range rather than a precise number, offering users a more flexible perception of wait times for ETAs above 6 minutes, which research has shown to be generally considered long by people in Indonesia. This approach helps reduce user anxiety by setting more realistic expectations and accommodating potential delays.

IMPACT

17% reduction in cancellations.
22% increase in user engagement.
Surpassing our target OKR by 2%.

The impact was measured using data collected two months after the public release of the new design enhancements in September 2021. The target OKR for reducing user cancellations was 15%, but thanks to these design initiatives, we achieved an 17% reduction— 2% more than what was expected!Moreover, this reduction in cancellations has saved Gojek a substantial amount in operational costs. While the exact figures are under NDA, the improved efficiency and user retention have contributed significantly to the company's bottom line.