Most HR systems hold far more than anyone reports on. The data is there, the dashboard exists, and the monthly review still runs on a spreadsheet someone rebuilds by hand.
The gap is not analytics. It is knowing which 6 numbers are worth watching.
Which numbers actually matter?
| Report | What it tells you | How often |
|---|---|---|
| Headcount by location and function | Where the company really sits | Monthly |
| Attrition, and inside 90 days | Hiring quality against onboarding quality | Monthly |
| Leave liability | A balance-sheet number, not an HR one | Quarterly |
| Absence pattern by team | Workload and management problems | Monthly |
| Statutory filing health | Whether PF, ESI and bonus went out on time | Monthly |
| Payroll corrections per cycle | Whether your data is trustworthy | Every cycle |
Six reports. If any needs a support ticket, the system is not answering questions, it is storing them.
Why is 90-day attrition the sharpest number?
Because it separates two problems people usually merge. Someone leaving at 2 years is a retention question. Someone leaving at 10 weeks is a hiring or onboarding question, and it is fixable.
Read it per manager, not per company. A company figure hides a problem that usually lives in one or two teams.
What makes leave liability a finance number?
Because unused encashable leave is money owed. An uncapped balance grows quietly until somebody resigns and the payout is larger than expected.
Pull it each quarter, by location, and read it next to the carry-forward cap. Balances above the cap mean the year-end rule did not run.
What does absence data actually show?
Patterns, not people. Monday and Friday bunching, one team always above the rest, or a site where sick leave spikes after a shift change.
For covered staff this ties to a legal benefit. The Employees State Insurance Corporation says "ESIC provides 70% of average daily wages in cash during medical leave, upto 91 days in two consecutive benefit periods." Absence is not only a cost line. It is a benefit entitlement with records behind it.
How do you measure statutory health?
By filing date, per stream, per month. PF at 12% with a 15,000 wage ceiling, held by the Employees Provident Fund Organisation. ESI at 3.25% employer and 0.75% employee up to 21,000. Statutory bonus between 8.33% and 20%, with an 8-month deadline.
Three streams, three dates. A single red cell on that report is worth more than any engagement dashboard.
Which data quality gaps should the report show?
- Employees with no UAN or partial KYC
- Nominations never filed
- Bank details that failed a payment
- Leavers with no exit date recorded
- Dependant records never updated after a life event
ESIC says "ESIC provides monthly payment apportioned among dependants in case of death due to employment injury", so the last line is not admin tidying. It decides whether a family can claim.
What does maternity cover add to the picture?
A planned absence you can plan for. ESIC says "ESIC provides 100% of average daily wages in cash up to 26 weeks in confinement and 6 week in case of miscarriage, during maternity leave and 12 weeks for commissioning mother and adopting mother."
That is 26 weeks you know about months ahead. A report that surfaces coming long absences by team turns a staffing shock into a plan.
What should you ignore?
Composite engagement scores with no action attached, vanity counts of logins, and any dashboard nobody has opened in a quarter.
A report is only worth running if a decision changes when the number moves. Everything else is noise.
How do you make reporting self-serve?
Insist during trial that you can pull 3 reports unaided. If the vendor runs them for you in the demo, you will be raising tickets for 3 years.
The employment framework sits with the Ministry of Labour and Employment. One employee record feeding every report is what complete HRMS and legal compliance are built around.