Standardizing Metric Definition and Process Management in Data Governance: A Ctrip Finance Case Study

Background – Since its establishment in 2017, Ctrip Finance has experienced rapid growth but faces data governance challenges due to frequent business iterations and a fragmented data architecture, leading to lagging data knowledge, redundant metric construction, and data quality risks.

Metric Definition Standardization – Inspired by the OneData methodology, the team unified metric definitions, data warehouse modeling, and development processes. They identified ambiguities in metric documentation and introduced a standardized workflow that includes unified metric entry, clear definition processes, and a four‑step metric definition flow illustrated with the example of daily iOS app registration users.

Systematic Process Management – To enforce the standardized workflow, a "Metric Standardization Management System" was built, abstracting the four workflow steps into modules: Business Domain & Data Domain, Business Process, Dimension Management, and Metric Design. Each module’s implementation is shown with diagrams and explains concepts such as business processes, transaction vs. snapshot processes, data‑table mapping, and dimension handling (derived attributes and linked dimensions).

Metric Design – The design distinguishes between atomic metrics (core aggregation logic) and derived metrics (group‑by and filter logic), using SQL to enforce consistent metric definitions. The relationship between business processes, atomic metrics, and derived metrics is summarized in a flow diagram.

Practical Application – After standardizing metric definitions, the team ensured application‑level consistency by linking each metric to a single task, moving from "requirement‑driven" to "metric‑driven" development. They introduced metric projects to group related metrics, aligned modeling with dimensional modeling theory, and facilitated business‑data research through the business process module.

Summary – The approach reduced data research costs, aligned metric definition with dimensional modeling, lowered data usage costs, eliminated metric redundancy, and improved development and operational efficiency.

Team Recruitment – Ctrip Finance’s data team is hiring Data Platform Engineers, Graph Storage & Computing Engineers, and BI Analysts. Interested candidates can email [email protected] with the subject "[Name] - Ctrip Finance - [Position]".

Recommended Reading

Million QPS, Millisecond Latency: Ctrip’s Real‑Time Big Data Foundation

How to Solve Real‑Time Data Aggregation

CrateDB in Ctrip Ticket BI Practice

Common Methods and Thoughts on Time‑Series Forecasting