Business Context

An Engineering Manager at NXP Semiconductors leads a 42-person firmware and validation team building software for the S32 Design Studio and supporting releases for the S32K automotive MCU platform. Over the last 2 quarters, leadership has seen mixed signals: release frequency improved from 6 to 9 production releases per quarter, but escaped defects rose from 14 to 23, and average cycle time increased from 11 to 16 days.

Metric Scenario

The VP of Engineering asks you to define a metric framework to determine whether the team is actually performing well, rather than relying on a single output metric. The team ships customer-facing SDK updates, internal tooling improvements, and silicon validation fixes. In the same period, story points completed increased 18%, on-time milestone delivery fell from 88% to 76%, Sev-1 production incidents increased from 2 to 5, employee attrition rose from 4% to 9%, and internal developer satisfaction dropped from 8.1 to 7.2 out of 10.

You need to recommend a balanced KPI set that captures delivery speed, quality, predictability, and team health, and explain how you would diagnose trade-offs between them.

Requirements

1. Define the primary KPI or scorecard you would use to evaluate engineering team performance.
2. Identify which metrics are leading vs lagging indicators.
3. Explain how you would decompose a decline in performance.
4. Recommend target ranges and guardrails so the team does not optimize one metric at the expense of another.
5. Describe how you would present this to NXP engineering leadership in a monthly review.

Data Available

• jira_issues: issue_id, team_id, issue_type, created_at, started_at, resolved_at, story_points, status
• gitlab_commits_prs: repo, engineer_id, commit_time, pr_open_time, pr_merge_time, review_cycles
• ci_cd_deployments: build_id, environment, deploy_time, release_version, rollback_flag
• defect_tracking: defect_id, severity, found_phase, opened_at, closed_at, product_area
• incident_log: incident_id, severity, start_time, end_time, root_cause, customer_impact
• eng_pulse_survey: engineer_id, month, satisfaction_score, burnout_risk, role_tenure