Most HR reports count what is easy: headcount, leave taken, courses delivered. Fine for a board slide. None of it tells you what to do on Monday.
Which numbers actually change a decision?
| Number | What it tells you |
|---|---|
| Exits inside 90 days, by manager | Hiring or 1 manager; both are fixable |
| Days to approve leave | Where the process stalls |
| Payroll errors per 100 payslips | Trust, measured |
| Leave not taken, by team | Workload and fear, before burnout |
| Promotion rate at the first manager rung | Who gets through, and who does not |
All 5 point at something you can change. Head count by team points at nothing.
Why do HR dashboards go stale?
They report the year. Decisions are made in a week. Drop any number nobody has acted on for 2 quarters.
Five numbers reviewed monthly beat 30 reviewed never.
What breaks the data underneath?
- The same person in 3 systems. Nothing adds up, so nobody trusts the total.
- Dates entered late. A joining date keyed a week after the fact ruins every tenure number.
- Free-text fields. 11 spellings of 1 department.
- No exit reason captured at the time. Added later, it is just a guess.
The second is the quiet one. Most firms do not know their 90-day attrition is wrong because the dates were entered in batches.
What should never be in a report?
Anything naming a person in a document that gets passed around. Aggregate to the team, and keep medical and conduct detail out of any shared view.
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." Someone claiming it has a medical history that does not belong in a dashboard.
How do you know a number is trustworthy?
It is computed from 1 source, the same way each month, and someone can say exactly which records it came from. If 2 people produce 2 answers, you have a wording problem, not a tool problem.
Write the definition down once. Most rows about data are rows about wording.
What does the outside picture look like?
The Ministry of Statistics and Programme Implementation says "Unemployment Rate (UR) in usual status (ps+ss) was 3.2% for both male and female of age 15 years and above during July 2023 to June 2024."
Outside numbers are context, not a target. Your own 90-day exit rate tells you more about next quarter than any national figure.
Does any of it connect to engagement?
Gallup says "The majority of the country's workforce (59%) consists of 'not engaged' employees", and Gallup puts manager engagement down from 39% in 2024 to 30% in 2025.
A number with no owner changes nothing. Give each of the 5 a name against it.
Where should the data live?
In 1 record per person, with dates entered on the day. Everything above fails if the underlying dates are approximate.
That single record is what 247HRM and performance management are built on, with PF records at the Employees Provident Fund Organisation and the framework at Ministry of Labour and Employment.