How JD Designs and Applies a Powerful Data Metric System

01 How to Understand Indicator Systems

This section introduces the basic concepts of indicators and indicator systems, illustrates them with everyday health metrics, and explains how business indicators are derived from data collection and analysis to objectively describe performance.

1. Basic Meaning of Indicators and Indicator Systems

Indicators are measurable values that quantify business processes; an indicator system organizes these metrics into a structured, classified framework that reflects the health of a business, similar to how medical check‑ups use height, weight, and blood tests.

2. Difference Between Indicators and Tags

Indicators measure business outcomes, while tags describe entity attributes. The differences are highlighted in three aspects:

Basic Meaning: Indicators quantify process and effect; tags provide comprehensive characterizations of users, products, or scenarios.

Processing Methods: Indicators are computed via statistical development; tags can be generated through statistical development, manual labeling, or algorithmic prediction.

Application Direction: Indicators support analysis such as monitoring and diagnosis; tags aid operational tasks like user segmentation and product selection.

3. Position of Indicator Systems in the Data Chain

Indicator systems sit at the Information layer of a four‑layer data pyramid (Data → Information → Knowledge → Wisdom), linking low‑level data collection to high‑level business insight.

They drive data‑layer modeling, improve data asset quality, and support business analysis by providing clear, traceable metrics.

4. Traffic Indicator System

Traffic metrics include exposure, clicks, PV, UV, orders, bounce rate, conversion rate, etc. Indicators are classified as atomic (directly aggregated from raw data) or derived (calculated from atomic metrics). Principles for building a traffic indicator system include scientific design, completeness, and business orientation.

5. How Indicator Systems Support Business Applications

The AIPL model (Awareness, Interest, Purchase, Loyalty) is used to quantify user journey stages, enabling precise operational actions such as brand exposure, targeted coupons, and repeat‑purchase incentives.

6. Understanding the Data Processing Logic Behind Indicators

Data collection starts with front‑end JavaScript that sends user interaction data to back‑end logs, which are parsed, cleaned, and stored in wide tables. From these tables, metrics such as UV, PV, and bounce rate are computed.

02 How to Build and Apply Indicator Systems

The OSM model (Objective, Strategy, Metric) guides the design of indicator systems by aligning metrics with business goals, processes, and strategies.

1. OSM Model for Indicator Design

Start with clear business objectives, decompose processes, and define metrics that support those objectives.

2. Deriving a North‑Star Metric

A north‑star metric should reflect core user value, activity level, be intuitive, decomposable, and support long‑term goals. Example: "Two hundred million users with an average of three uses per year".

3. Combining OSM with Business Strategy

Map business goals to workflows, then to data flows and metrics (e.g., efficiency and experience metrics in JD logistics).

4. DuPont Analysis for Goal Decomposition

Break down profit into revenue, margin, cost, and further decompose revenue into order volume and average order value, linking each to underlying indicators such as conversion rate, UV, and user frequency.

03 Key Guarantees for Effective Indicator System Landing

Standardized indicator definitions and development practices ensure consistency and reliability.

1. Indicator Definition Standards

Use a 4W1H + Dimension framework: What (business process), Who (entity), How much (measure), Where (data scope), Dim (dimension combination), When (time range). Example: fin_ob_order_amt for outbound order amount.

2. Indicator Development Standards

Adopt low‑code platforms (e.g., JD Logistics Udata) to let non‑technical users define, configure, and generate indicators, ensuring uniform definitions and efficient development.