Most HR dashboards report headcount, attrition and a few training numbers. None of those, on their own, tells anyone what to do differently. A metric earns its place by changing a decision.
Attrition, properly segmented
Overall attrition is nearly useless. 18% could be entirely healthy or a crisis depending on who left. Segment by:
- Voluntary versus involuntary — these are different problems with different owners
- Tenure band — attrition in the first 12 months indicts hiring and onboarding; attrition at 2–4 years indicts career paths and compensation
- Performance rating — losing low performers is housekeeping; losing high performers is urgent
- Team and manager — attrition concentrated under one manager is a management issue, not an HR one
- Reason, from exit interviews, read sceptically since people soften their answers
Absence patterns
- Absence rate — unplanned absence days as a percentage of available working days
- Frequency versus duration — many short absences indicate something different from one long one
- Monday and Friday concentration — a known engagement signal
- Leave not taken — employees with large untaken balances are a burnout and liability risk simultaneously
- By team — concentration points to workload or management
Cost per employee
Total people cost — salary, statutory contributions, benefits, recruitment, training, HR function cost — divided by headcount. Useful for budgeting and for understanding the real cost of attrition. Track the trend rather than the absolute, and compare across business units.
Internal mobility rate
Internal moves and promotions as a percentage of total roles filled. Low internal mobility predicts mid-tenure attrition reliably: people who cannot see a next step leave to find one. It is also a leading indicator, which makes it more useful than attrition itself.
Time to productivity
How long until a new joiner performs independently, as assessed by their manager. Measures hiring quality and onboarding effectiveness together. Track by role family and by source — it often reveals that your cheapest hiring channel is the most expensive once ramp time is counted.
Span of control and manager load
Direct reports per manager. Both extremes are problems: too many and managers cannot support people, too few and you have excess management layers. Combine with attrition by manager for a sharper picture.
Payroll accuracy
Errors per thousand payslips, and corrections required after finalisation. A rarely tracked metric that correlates directly with employee trust. If you have never measured it, the first month is usually sobering.
Compliance health
- Statutory filings submitted on time, as a percentage
- Penalties and interest paid
- Open audit findings, by age
- Documentation completeness — employees with missing PAN, UAN, nomination forms
Metrics to retire
- Training hours delivered — measures activity, not capability gained
- Total headcount — context, not a metric
- Engagement survey score in isolation — a single number with no segmentation hides everything useful inside it
- HR-to-employee ratio — benchmarked endlessly, rarely actionable
Presenting metrics so they get used
- Show trend, not snapshot — three months minimum
- Segment by team and tenure as standard
- State what you think it means, and what you propose doing
- Limit any one report to five or six metrics
- Send to the people who can act, not only to leadership
