1- Didi Premium Service Internship.png

Didi Premium Service

Hello world

1 - Hero-img - 2.png

Didi Premium Service Internship

App Design, Ride-sharing, AI


Helping Didi premium service achieve 30% of market share

As a UX design intern at Didi, I helped Improved the Premium ride hailing experience at Didi. Worked with marketing, product, loyalty member management and big data teams. Researched premium user persona, designed an end-to-end premium user experience that helped Didi win market shares in the premium user segment and improved loyalty user's stickiness.

Length of the project: 3 months
My role: User research, User experience design, Visual design, Prototyping
Teammates: Xinrui Li (Design lead), Yang Sun (Product designer), Aya Wang (Product designer)
Collaborating departments: Product Management, Data Science, Marketing, Customer Service
Keywords: Mobile App, Artificial Intelligence

To comply with my non-disclosure agreement, I have omitted and obfuscated confidential information in this case study. All information in this case study is my own and does not necessarily reflect the views of Didi.

Project impact

2 months after launch, daily premium users increased from 650K to 4 million, capturing 26% of the market


Design process


Problem statement

Didi,  the biggest ride-hailing and sharing digital player in China, accounted for 90% ride hailing market with over 400 million riders and 20 million drivers. Since 2017, Didi began to enter premium ride market and released two premium ride services:  Premier and Luxe. After 6 months’ effort, Didi only won 5% market share, far from its objective in 2017: 30% share in premium service market. To quickly change the weak market position, Didi requested its Premium Service UX team to work across departments to achieve Didi’s business goals.

How can Premium Service UX team help Didi improve its market share in premium services to 30% by the end of 2017?

Problem statement - Pie chart.png

Final Deliverables


1.Identify and convert potential premium users


Persona summary

Existing Premium service users but don’t use Didi’s premium services.


00.Identify premium users

We used the existing Didi data to identify the potential premium users. We used frequency of rides, the average spending on each ride, and the number of rides per month as criteria for our machine learning algorithm to search for potential premium users.

01.End-of-trip promotional incentives

When a potential premium user, Sarah, is identified, Didi sends her in-app messages that she can enjoy 50% off discount on her first premium Didi service ride. When Sarah uses Didi over the weekend, the message touches her, and she proceeds to try out Didi premium service over the week days.

02.Before-the-trip promotional incentives

For potential premium users who are less likely to convert, if they perform actions that indicate a higher chance that they are likely to convert to premium users, for example, when they set destinations at places where premium service users frequent, Didi will send them in-app messages to motivate them to use premium ride services


03. During-the-trip reservation reminders

When potential premium users set destinations at commute locations, Didi reminds them to reserve a return trip with the premium ride services during their trips.


2.Increase premium users’ loyalty

Persona summary

Loyal premium users who are not satisfied with Didi’s existing loyal user benefits.



In China, during peak hours, the ratio of ride-hailers to vehicles often is at a 20:1 or even higher. Therefore, many premium users often are frustrated by the long wait time.

Expressway helps loyal premium users skip the queue and get the ride under 5mins

2 - Expressway.png

02.Free upgrade

When loyal premium users hail economic ride options, Didi upgrades their economic ride options with additional services for free. Free-upgrade incentivizes loyal premium users to continue using Didi’s products over the week days



Request multiple rides is designed for users who are very flexible about the cost of their ride options, such as Amy. With Request Multiple Rides, Amy can select multiple rides and request them altogether. Didi will match Amy with the vehicle closest to her.


3.Improve Didi’s service qualities

Persona summary

Past premium service users. They stopped using Didi’s services because they have had bad ride experiences.


01.Feedback Redesign

Changed to 2 options to reduce users’ cognitive load and help Didi improve net promoter score:
1.Detractors - Angry
2.Passives - Okay
3.Net promoters - Super satisfied


Principle Prototypes



1.Conducted competitive analysis of Didi’s competitors to understand the current landscape of premium service market.
2.Conducted focus groups with 50 premium service users and in-app surveyed over 1000 Didi users. to understand who are the premium service users and why they don’t use Didi’s services
3.To learn more about the research process, click on the button below


Research Insights

Insight 1

90% of the premium users in China are already Didi users, they simply don’t use Didi’s premium services


Insight 2

3 main reasons that they don’t use Didi’s premium services are that they don’t know that Didi has premium services, they are unsatisfied with Didi’s premium service offerings and they are unsatisfied with Didi’s premium service qualities


Research Synthesis Artifacts


Based on the insights created from research synthesis, I created three personas that capture the different characteristics of premium ride service users in China


Customer Journey Map

I created Customer journey map to both visualize users’ end-to-end experience and define opportunities we have to address the problems discovered 




Based on the customer journey map, we brainstormed a series of features that could fulfill user needs. We used value proposition map to filter the features that both align with our business objectives and fulfill users’ needs:

1.Increase awareness & add competitive incentives

Didi’s AI identifies potential premium customers and push notifies potential premium users that their first Premium rides are 40% off

2.Additional benefits for loyal users

Add top features requested by existing premium users:
2.Free upgrade
3.Request multiple vehicles at once

3.Improve premium service qualities

Improve feedback at the end of the trip to gather actionable feedback on the performance of drivers, vehicles and Didi.

15 • Brainstorm Diagram.png



For each feature, I identified primary use cases and created user flows to help my team understand both the front end user interactions and backend Didi system interactions. The front-end facing interactions illustrates steps users take to achieve their goals and backend interaction helps uncover data Didi app needs to query from Didi’s database.Below, I demonstrate the design solutions that I came up for 3 parts of user’s journey: 1.New User Acquisition 2.Loyal User Retention 3. End-of trip Feedback


1.Acquire Potential Premium Users

Persona #1: Sarah, a potential premium user of Didi. She is a premium ride service user but she hasn’t discovered Didi’s premium services yet. Even if she knows that Didi has premium services, she needs attractive new user incentives to influence her to use Didi’s premium service. 

00.Identify premium users

Based on Sarah’s Didi usage data, Didi AI tagged Sarah as a potential premium user. It then proceeds to send Sarah in-app messages on her new user incentives on using Didi premium service.


01.Promotional incentives

The promotion incentives appear in-app when Sarah requests a ride, during the trip and at the of her trip with Didi. 


2.Increase loyal users’ stickiness

Persona #2: Sarah, Diamond member Didi Premium user (2nd most valued user tier). As a Diamond member, she enjoys both Expressway and Free Upgrade as value-added services that are exclusive to her membership tier.


When a Sarah enters a waiting queue during peak hours, Didi provides her Expressway to skip the long line and request her ride first. After invoking the Expressway, Sarah becomes the 2nd in the line to hail a ride and saves 30mins of wait time.

However, Didi Expressway is disabled when these conditions are not met: 

1.More than 10 users hailing the same service at the same time, and the number of Didi rides within 2 miles is less than 1/3 the number of users hailing vehicles. 

2.There are less than 5 loyalty users using Expressway at the moment

3.The user still has remaining Expressway left for the quarter (each Diamond / Black Gold level user has 3 Expressways per quarter)

There are 5 use cases for Expressway:


02.Free upgrade

From our research, we found that Sarah uses economic Didi services for their personal use and family trips over the weekend (she pays these rides out of her pocket). When she hails economic ride options, she has a 75% chance to receive an economic ride service with top additional services free-of-charge. 

These free upgrades incentivize Sarah to continue using Didi’s premium services so that she can enjoy high quality economic services for personal and family rides.



Amy, during the peak hours, selected four rides to request. In a few seconds, Didi matches her with Didi Pool. Previously, she would have to tap over 12 times to get a vehicle. The new design’s interactions take 3 steps.


3.Improve Didi’s service qualities

01.Feedback Redesign

Persona #4: Vincent, a Didi loyal premium user. He loves the affordable price that Didi premium service offers but he’s not satisfied with the professional performance of Didi’s premium service.

With the new feedback system, Vincent can provide detailed feedback on which part of the service to improve: Driver, Vehicle and Didi.


User testing


01. in-person user-testings

We conducted first round of in-person user-testings with 4 groups of Didi users. 2 groups of 10 existing premium users and 2 groups of 10 potential premium users. The user-testing is conducted via think aloud.

Key findings:

1.Instead of providing a message on the discounted vehicle, provide users a quick entry to the discounted vehicle because users might not be familiar with the service and know where to find it
2.Users would like to customize the ride they would like to request to gather based on their individual preferences

02: Soft-launch

Soft-launch the features to 10,000 premium users.

Key findings:

1.The conversion rate on Expressway, Request multiple rides are around 92% and 74% respectively.
2.Premium users live relatively close to each other. Therefore, often more than 10 premium users are in Expressway together, inhibiting the quality of the Expressway.
3.Premium users would love to pay or trade mileages for more Expressways each month


Reflections & Future Thoughts


Future improvements

Features that we launched so far still focus on maintaining the existing premium users on the platform. However, in order to achieve the business goal of expanding to 30% of market share, potential premium user acquisition features need to be closely monitored and analyzed for its efficiency.

Continue monitoring the data collected after officially launching the feature on Didi platform. Continue making refinements to the features based on collected users’ data.

Thoughts about VUI

Another opportunity to improve the existing ride-hailing experience is to introduce voice interactions to the platform. Voice interactions will not only make the ride-hailing experience more natural and seamless. The vehicle matching process is completely left to machines, which can more effectively pair users with ride than users selecting which ride they want manually. 

Thoughts about AI

My proposed features involve using Artificial Intelligence. The beauty of AI and any other cutting edge technology is that they can provide customized services to users by identifying user type and their needs (e.g. when Didi AI identifies premium users, Didi AI recommends them to book rides for the week ahead of time because premium users commute). Still, with access to these personal data help us bring convenience to individuals while at the same time giving Didi the ethical and social responsibilities to protect these data.