Application of Data Tags and Metrics in the Financial Industry

This article explains the concepts, classifications, construction methods, and practical usage of data tags and metrics in the financial sector, illustrating how to build indicator and label systems and how to apply them effectively for refined customer operations and business management.

01. What are metrics and tags, and what is the difference?

Metrics usually describe objective facts and are often numeric, such as GDP, CPI, or financial indicators like loan volume or delinquency rates. Tags are human‑defined categories (e.g., high‑net‑worth customers, product risk levels) that combine objective facts with subjective definitions to group entities.

Metrics can be derived from tags (e.g., migration rate from one tag to another) and tags can be derived from metrics (e.g., defining a "private banking customer" as AUM ≥ 5 million). Thus, the two can be transformed into each other.

Metric classification

Metrics are typically divided into atomic metrics (directly counted, e.g., customer count), derived metrics (aggregated from atomic metrics with dimensions, e.g., mobile‑banking customer count), and ratio metrics (derived from other metrics).

Tag classification

Tags include fact tags (directly from raw data, e.g., gender), rule tags (based on statistical rules, e.g., age groups), and model tags (abstracted from fact and rule tags, e.g., "moonlighter" or "white‑collar").

Metric usage scenarios

Metrics are usually broken down for monitoring and evaluating business performance, such as splitting total customer count into dimensions like channel or time period.

Tag usage scenarios

Tags are used for classification and targeted operations, such as segmenting customers for precise marketing, product recommendation, or eligibility assessment.

02. How to build a metric system?

The metric system should combine top‑down (strategic objectives broken into primary indicators) and bottom‑up (operational metrics collected from frontline staff) approaches. Example: a bank’s strategic goal of “high‑efficiency operation” is decomposed into key factors, forming a multi‑layered framework.

Top‑down steps include defining a few North‑Star metrics, then decomposing them into sub‑metrics across business modules. Bottom‑up steps involve interviewing staff, collecting frequently used operational metrics, and merging them with the top‑down set to form a complete system.

Metrics must be standardized with business, technical, and management attributes (identifier, name, definition, data type, calculation logic, responsible department, etc.) and visualized via dashboards or reports.

03. How to build a tag system?

The tag system should align with business goals, defining tag names, meanings, and scopes similarly to metrics. It requires clear identification of which business value layer each tag belongs to, enabling targeted supplementation where tags are missing.

04. How to use tags and metrics effectively?

Common issues with tags include mismatched expected vs. actual customer counts and duplicate processing; solutions involve verifying calculation logic and monitoring tag usage frequency. For metrics, break down large indicators into sub‑indicators to pinpoint root causes (e.g., distinguishing between external economic factors and internal approval processes).

Summary

The article first distinguishes tags and metrics, then describes how to construct metric and tag systems—combining top‑down and bottom‑up methods for metrics and business‑driven design for tags—and finally explains best practices for applying both, emphasizing lifecycle management and operational integration.

Q&A

Q1: How to effectively identify events? Collect user‑behavior events, track high‑frequency events with response rates after applying strategies, and retain those that show measurable impact.

Q2: Recommended books on tag construction? "Tag Category System" and "User Portraits: Methodology and Engineering Solutions".

Q3: Should metrics and tags be built in separate systems? Yes, because their usage scenarios differ: metrics support KPI tracking, while tags support customer segmentation and marketing.