Meta VO Interview Question: Carpooling Value Proposition, Metrics, and Data Model Design
A ride-share service allows customers to use an app to request a driver to pick the customer up and drop them off at their destination. To reduce costs and the frequency of required trips, customers can select a carpool option, which will overlap their route with other customers, resulting in an overall longer trip with multiple pick-ups and drop-offs.
Common considerations when a carpooling algorithm is deployed:
- Type of vehicle and maximum occupancy
- Historical demand of trips along the original route
- Customer ratings
- Value proposition to customer (how much is carpooling incentivized)
Let’s say we recently launched this carpool option in the last couple of months. The executives want to know if the option is increasing app usage, and if the value proposition passed along to the customer is high enough to encourage usage, without eating into the company’s profits.
Q1. What business questions would help us understand the value proposition offered by the carpooling option?
Q2. What metrics would you define to help answer the business questions from Q1?
Q3. What are the most important dimensions/cuts that would be relevant for your metrics and why?
We would like to design a data model to be able to answer our business questions, such as:
- How often is carpooling selected compared to regular rides?
- What is the average amount saved by the company through carpooling?
- What is the average cost passed on to the customer?
With this data model, we wish to track:
- Trips with carpool vs. all rides
- Costs saved by carpooling vs. price incentives to customers
Q1. What would be the key entities and attributes for each entity in this model?
Q2. How would you design the data model to support the entities and attributes? What are the key tables you would build and what type of relationship exists between them (1:1, 1:M, M:M)?
这道 Meta VO 题目围绕网约车拼车（carpool）功能展开，核心不是写算法，而是做业务分析与数仓 / 数据模型设计。第一部分要求先定义业务问题与指标，比如拼车选择率、拼车带来的增量使用、公司节省的成本、向用户让利的平均幅度，以及这些指标在不同维度下的切分方式，例如时间、城市、车型、乘客类型、是否新老用户等。第二部分进一步要求抽象实体与表结构，通常会围绕 trips、customers、drivers、vehicles、carpool offers/discounts 等核心对象建模，并说明它们之间的 1:1、1:M、M:M 关系。解题重点是把“拼车是否提升使用、是否影响利润、是否真正传递价值”转化为可度量、可落地的数据模型。
