M2-012Hard市场与生态 · Marketplace & ecosystem
一个双边市场产品的“流动性”下降了,你怎么诊断?
Liquidity in a two-sided marketplace is falling. How do you diagnose it?
💡 考场高分一句话答案:
把流动性拆成需求侧、供给侧、匹配效率三部分,再按地域/品类/时段分层,判断是供给撤离、需求变化还是匹配变差。
Split liquidity into demand, supply and matching efficiency, then segment by city, category and time of day to see whether supply withdrew, demand changed, or matching got worse.
📌 核心考察要点 (Key Points)
- 需求侧:请求量、搜索量、下单意图(是否有需求但没被满足)
- 供给侧:可用供给数、在线时长、接单率、供给方留存
- 匹配效率:fill rate、time-to-match、取消率、匹配距离/价格分布
- 分层:按城市/品类/时段——流动性问题几乎总是局部的(某些城市/时段先坏)
- Demand side: requests, searches, purchase intent — is there demand that goes unserved?
- Supply side: available supply, online hours, acceptance rate, supplier retention
- Matching: fill rate, time-to-match, cancellation rate, distribution of match distance or price
- Segment: by city, category and hour — liquidity problems are almost always local and start in a few cells
📐 第一性原理与数理推导机理
流动性可以写成 $L=\frac{\text{matched}}{\text{requested}}=f(\text{supply density},\ \text{demand volatility},\ \text{matching algorithm})$。供给密度下降与需求波动上升都会降低 $L$,但解法完全不同:前者要补贴供给,后者要平滑需求(预约、动态定价)。
⚖️ 业务权衡与取舍 (Trade-offs & Calibration)
双边市场诊断的关键是先看局部。全局 fill rate 从 85% 降到 83% 可能只是几个城市崩了,其他城市没变。因此标准做法是画“城市 × 时间”的流动性热力图,找异常格子。这也是 marketplace 类面试的高分动作。
⚠️ 常见高频面试陷阱与避坑指南
- 只看全局指标,不做地域/时段分层
- 把流动性下降直接归因于算法,不先看供给量
🔍 大厂面试官连环追问 (Follow-ups)
- 如何区分“供给不足”与“匹配算法变差”?
- 动态定价会如何影响流动性指标的可比性?
嫌字多看太累?进入苏格拉底式对话推演实验室
零打字负担,以大厂考官真实困境切入,通过点选直觉选项暴露思维盲区,手把手推导 Why。
⚡
Test muscle memory on this topic via 3-Layer Flashcards
Active Recall: Mentally formulate Prompt ➔ Math Derivation ➔ Benchmark Implementation in 3 minutes.