Mastering Indicator Systems: How to Design, Manage, and Leverage Business Metrics

01 How to Understand Indicator Systems

Even though internet companies are highly digitalized, clearly seeing a company's operational status remains difficult; it involves both investment and business performance visibility.

Key questions include: how many users are gained from a multi‑million budget, conversion at each stage, 1/7/30‑day retention, and channel quality evaluation.

In data‑driven companies, decisions are made only with data, requiring mature dashboards; ad‑hoc requests cannot be handled without a solid indicator system.

An indicator system is simply "statistical results that help the business"—they describe what happened, measure how much, and break down the reasons.

Indicators can be classified as atomic (e.g., PV, UV, order count) or derived (e.g., average transaction amount, conversion rate, recent N‑day order volume).

02 How to Design Indicator Analysis

Designing an indicator system follows three stages: 1) define business goals, 2) build an analysis model, 3) collect, aggregate, and display data.

The system must align with senior strategic goals; indicators are tightly linked to KPIs and drive performance improvement.

Examples: for e‑commerce, the goal to increase revenue leads to focusing on order volume; for enterprise SaaS, the goal to acquire more leads leads to tracking registered users.

Common analysis models include OSM , PLC , and AARRR . OSM breaks a big goal into actions:

O – Objective: what does the user want to achieve?

S – Strategy: what strategy will achieve the objective?

M – Measurement: what metrics reflect the strategy’s impact?

In cloud computing, an OSM breakdown might look like the diagram below.

The PLC model divides a product lifecycle into Exploration, Growth, Maturity, and Decline, each with focus metrics such as PV/UV, retention, DAU, and churn.

Other models like AARRR or RFM are also used but not detailed here.

After defining the analysis goal, data must be collected and visualized, often involving data modeling and dashboard construction.

03 Indicator Management Methods

An indicator system requires ongoing management and interpretation.

Example: during a promotion, operations want to calculate "hot‑item rate" (items selling >20 units divided by total items). Pitfalls include unclear definitions of "sales" (cart adds, orders, payments), handling refunds, SKU granularity, dimension and period of aggregation, and the metric’s audience.

Mis‑defined metrics cause data rework, disputes, and complaints.

Another example: a ride‑hailing service sees a 50% revenue drop; analysis reveals a 20% drop in order volume due to pandemic, but the remaining 30% is unexplained because the indicator system lacked relevant metrics, leading to urgent ad‑hoc requests.

Not every metric has real business value, yet each requires dedicated resources. Teams must evaluate the cost‑benefit of computing each indicator, considering storage and compute expenses.

This aligns with Alibaba’s OneData methodology: unify definitions, reduce disagreements, accurately measure impact, and lower cost.

04 Correctly Recognizing Indicator Systems

Many seek a one‑size‑fits‑all template for indicator systems, but each business has unique metric needs; templates can only serve as shortcuts for mature domains like e‑commerce.

The difficulty lies in management, not technology. As industries mature, analysis methods become standardized, and indicator systems fulfill their historical mission.

When a market reaches saturation, new opportunities emerge, and the era shifts from "everyone is a product manager" to "everyone is a data analyst".