didi

"Work efficiency is the key."

Running a huge e-commerce business is complex. An efficient
internal product is needed to handle the fast-changing business.

My role

Product Manager
Worked with other PMs & Engineers

Responsibilities

Product Strategy.
User research.
Interaction design.

Duration

2021/08 - 2021/09
2 months full-time internship

Didi is one of the world's largest ride-hailing companies. In order to expand its business and get into different sectors, it launched "Chengxin Youxuan" in 2020 —— a community group buying platform in China.

As the e-commerce business was growing and became quite complex, a better internal product to improve work efficiency is needed. We want to facilitate users (in this case, purchasing specialists and operations) to locate "bad" products & remove them from the platform.
‍
Due to the internship agreement, I can't disclose any internal interface and data. So for this project, I will only present my responsibilities and design methods in general.
‍

Understanding the business

What's for sale?

It sells all kinds of products, mainly grocery. All the products have been divided into 4 categories:
① Fruits & vegetables ② Meat, Dairy & Eggs ③ Standardized food products ④ Standardized non-food products

Where does the service cover?

The service covered most major cities in China. All the area was divide into 5 main districts for more convenient management. Each main district has its own subdivisions.
‍

How to put products for sale & remove products from the platform?

The illustration below shows the process of putting products for sale on the platform.
‍

Note: For some areas, the Purchasing Specialist & Operations may be the same group of people. The job boundaries may not be that clear.

The illustration below shows the process of removing products from the platform (no longer for sale).

What's my responsibility?

I was responsible for redesigning the products removal (no longer for sale on the platform) system. My goal is to create a reasonable & effective system to help users locate "inefficient"/"bad" products effciently.

The process that I focused on is shown below. The target users are: Purchasing Specialist/Manager & Operations (nationwide).

‍
My job included:
① User research ② Brainstorm new rules with other PMs ③ Create prototypes for the change
③ Write PRD & go through internal critique ④ Worked with engineers to implement the iteration

Why products removal is a need?

Before going into the research, the first thing I tried to understand is why we need to do product removal. What is the significance and purpose of it?

The platform had thousands of products for sale in each city. We want customers to find everything they need on the platform (abundance of products) while only put the "best"/"efficient" products for sale. As you can see above, the homepage of the app (the first image) provides lots of information. And there was limited space to showcase products.

If customers are not interested & not even click into the product details / receive the product but not satisfied with it, there will be less purchase. It is important to only show "efficient" products that lead to purchase, otherwise it would be a waste of product impression.

‍So removing "inefficient" products (combined with adding good products) could
‍① Improve platform branding by only providing "good" products ② Increase conversion rate in general

What's the current problem of locating "inefficient" products?

Since each district was doing products removal in their own way, what's the need of creating a universal system to facilitate the process?

Doing products removal could be time-consuming and unorganized. Our users (purchasing specialists & ops) need supports and an effective system from the headquarter to improve the efficiency.
‍

What's the problem of previous beta system?

The previous beta testing system focused only on GMV (Gross Merchandise Value) ranking and the removal cycle was 7 days.

There were various factors that influence GMV (e.g. not enough exposure, price, seasonal reasons), and GMV shouldn't be the only standard to determine whether a product is "good" or "bad".
‍
‍
‍

Possible definition of "inefficient" products

I brainstormed with other PMs about taking potential factors (besides GMV) into considerations.

① High complaint rate / Few repurchase

Even if the GMV of a product is high, it does not ensure its quality. In order to keep a nice branding of the platform, poor-quality products should be removed.
‍

② Few users number

One big standard that determines whether a product is valuable is: how many unique users have bought it.
‍

③ Unstable supply

Since the products were provided and delivered by the supplier, there were circumstances that some suppliers failed to deliver the products on time, which made the products constantly out of stock. It increased the management costs.
‍
‍

User Research

We interviewed 24 users online, including purchasing specialist/manager & operations from 5 different districts and 4 product categories, and tried to get feedback about the previous beta system & figure out what's the best way to build a system to help.

‍
Each online interview lasted for about 20 minutes. The goal of the interview was to see how they were doing products removal on their own (including removal rules, cycle, special conditions), and tried to draw a conclusion and brainstormed for the new removal system.
‍

- Key finding 1: Various considerations for "inefficient" products (mostly GMV)

Most users took GMV as the major factor, but they also took various factors into considerations, such as user number, complaint rate, and the category products belong to.
* For example, due to the nature of some standardized non-food products like tissue paper, the sales in a short period will be lower than other grocery like meat & vegetables.
‍

- Key finding 2: Need time to take actions on products with potential

When the products' sales are poor, purchasing specialists/operations could take actions to "save" the products that they believe are "efficient" (e.g. increase exposure, lower the price...) instead of directly removing them.
‍

- Key finding 3: Similar removal cycle length

Nearly all the users complained about the short removal cycle of the previous system (7 days). Most users conducted products removal twice/once a month, no matter what category the products belong to.
‍

Research conclusion

After the research, we soon realized it was too complex to design a meticulous system tailored to each product category. So the best way is to create a rather flexible system.

Design solutions

- Add user number into the system

Besides GMV, we added user number as a new factor when doing removal. If the GMV and user number both ranked under a certain value, the product falls into the removal pool. We understand it would be more effective if we take every factors into considerations, but it is not practical to make such a big change for one iteration.
‍

- Extend the removal cycle

We extended the previous removal cycle time to 21 days. The longer cycle should be more accurate to assess products' "efficiency".
‍

- Add removal warning

In the 21-day removal cycle, a products list based on the ranking of GMV & user number for the first 14 days will be given to the users as product removal warning. Users have 7 days to take necessary actions to save some "good" products from the warning list. If the rank of the product is still under a certain value in the 21 days, the products will be officially added into the removal list.
‍

What I Learned

This is my first to-B product experience. The biggest challenge was to understand the business (and this is a quite complex one, only parts of the business is presented). Working in a big and fast-paced company was also a valuable experience, it was extremely important to improve efficiency by keeping controls of meetings. It also improved my user research skills —— gathered information and drew design solutions.

The experience definitely helped me understand the logics behind e-commerce, and also taught me what to think about & what's the priority when building a product —— always ask the fundamental questions first.
‍
‍

→ Next: Job Referral Platform Design