Internet operation should be user-centric, and users are the real bosses in the Internet era! With the continuous accumulation and precipitation of users, when users reach a certain level (there is no fixed value), it is very important to define the rating for their own users.
If we divide the company's customers into four quadrants according to the two dimensions of order frequency and customer unit price, and divide them into five groups: A, B, C, D, and E, one of the core demands of the enterprise is to find more potential customer group A, convert it into customer group B, and strive to make customer group B transfer to customer group E, and try to stay in the position of E.
Past Questions: NetEase Written Exam
Koala Overseas Shopping has always been user-centric, providing users with high-quality products to help users "live a better life with less money". In order to meet the needs of different users (for example, the requirements of new customers may be different from those of old customers, and churned customers need special care), please design a set of specific solutions, reasonably divide different users, and give corresponding suggestions.
Answer:
(1) To divide customers, it is to find two coordinate axes. During the interview process: we can analyze the life cycle of the customer. Then pick the two most important indicators.
Generally, the Y-axis is "customer value"!
X-axis: According to the meaning of the question, you can choose "churn rate" for this question.
So how do you calculate the x and y axis indicators? Using the RFM model: R (Recency): The interval between the customer's last trading hours. F (Frequency): The number of transactions made by the client in the most recent period. M (Monetary): The amount of money that the customer has traded in the last period of time. ）
(2) Countermeasures: How to maintain loyal customers or recover lost customers?
First, analyze the reasons for customer churn: price, season, channel, competitors. Use A/B testing to find the "G-spot" that users are comfortable with: expose a subset of people to a "wow moment" during the experiment and see how it affects retention.
Then do some corresponding activities: (1) exclusive activities on member days, which can stimulate users to consume and obtain higher privileges through consumption (2) point coupons, points redemption, points exchange, etc
Past question: NetEase interview
HOW DO YOU BUILD A MODEL THAT PREDICTS THE CHURN OF AN APP? What features can be created?
Answer:
Define the concept of customer churn with the "RFM model" judgment.
In the Analysis Window (January to December), the number of users on consecutive inactive months from 0 to 12 is counted.
Calculate churn probability:
JD Interview Questions
How to choose the best ten out of 100 merchants
How do you estimate how many barbershops there are in a city?
Answer:
Starting from demand, the number of barbershops = total population * how many haircuts per year divided by the number of people a shop can serve in a year
Koala Overseas Shopping Interview Questions
Always user-centric, provide users with high-quality products, and help users "live a better life with less money". In order to meet the needs of different users (for example, the requirements of new customers may be different from those of old customers, and churned customers need special care), please design a set of specific solutions, reasonably divide different users, and give corresponding suggestions.
Answer:
Users are divided into the following three dimensions, that is, each user will be tagged with the following three labels
1. New and old customer labels
Whether the user's device ID appears for the first time to distinguish whether the user is a new customer or an old customer, for new customers, new customer guidance can be carried out and new customer first order preferential activities can be provided
2. Crowd tags
Enumeration values: male, post-70s and pre-70s female, post-80s female, post-90s female
The data of this tag is derived from the user's personal information and behavior
Major groups of people have different preferences for brands and categories, and personalized recommendations can be made, such as recommending men's clothing and sports brands to men.
Recommend home furnishing categories to post-70s women, recommend maternal and child products to post-80s women, recommend cosmetics to post-90s women, etc., of course, this operating group tag can be refined according to the user's browsing/purchase history, if there are female users who have browsed or purchased maternal and infant categories, they can be classified as post-80s women-married-have children
3. Active status tags
Determine the active status of the user according to the user's purchase, such as an order within 30 days is regarded as an active user, and an order within 80 days is a high-risk user.
There are no orders within 160 days, and there are historical orders for churned users, etc
Enumerated values: Active, High-Risk, Sleepy, Churned, Registered Not Purchased
For active and high-risk groups, preferred brands can be recommended based on their behaviors to stimulate consumption
For sleepy and churned users, you can push or SMS push coupons and event previews
For users who have registered but not purchased, relevant prompts can be made, such as information such as the purchase of the first order discount
- End -