Quality Operations for Intelligent Payment: Improving Test Phase Metrics

Background: Quality is critical for product success. In fast‑iteration internet companies, balancing speed and quality is a key QA challenge.

Challenge: Meituan Dianping Intelligent Payment processes over 10 million daily transactions. Rapid growth amplifies quality issues, increasing resolution cost.

Pain points identified during test phase:

High‑severity defects cause frequent test rejections.

Large defect volume leads to long locate‑fix‑regress cycles.

Low‑level defects (typos, variable errors) erode team trust.

Key difficulties:

Achieving shared understanding of reported quality problems.

Quickly analyzing and locating root causes across requirements, design, development.

Deriving actionable improvement measures from generic suggestions.

Approach: Apply quality‑operation methodology (PDCA) focusing on test‑phase metrics: defect count, severity, and origin.

Solution highlights:

Align goals with QA KPI and stakeholders.

Present objective defect data (counts, severity distribution, cause breakdown).

Enable flexible data drilling across dimensions.

Implement concrete, review‑driven improvement actions and automate checks.

Metrics defined and standards set (examples):

Defect rate per KLOC: Mobile < 0.45, Frontend < 0.2, Backend < 0.15.

Sonar severe issues per KLOC: Blocker = 0, total < 0.1.

Severe defect ratio < 3.5%.

Requirement defect ratio < 10%.

Data is collected via the internal Metrics platform, with secondary processing for accurate calculations.

Improvement cycle combines top‑down (leadership‑driven data awareness) and bottom‑up (case review, data drilling) to close gaps.

Multi‑dimensional aggregation (weekly, monthly, quarterly) supports risk warning, confirmation, and assessment.

Standardization and tooling: automate test‑gate checks, adopt “poka‑yoke” mechanisms to prevent non‑conforming outputs.

Results: Overall metrics meet targets, though some areas (e.g., mobile defect rate) still above thresholds. Quality data usage increased across QA, developers, and leaders, and iteration efficiency improved.

Conclusion: Continuous quality‑operation improves test‑phase quality, promotes a quality‑centric culture, and supports rapid business growth.