Feature lists are the least useful part of choosing an HRMS. Every vendor has 200 of them. The demo shows the 20 that work best. What matters is whether five things hold up under your real statutory and working load.
Does the payroll engine handle Indian statute properly?
This is the one that cannot be worked around. PF runs at 12% with a 15,000 wage ceiling. ESI is 3.25% employer and 0.75% employee up to 21,000. Professional tax differs by state. TDS now carries a regime election. Statutory bonus sits between 8.33% and 20%.
The rules sit with the Employees Provident Fund Organisation and the Employees State Insurance Corporation, which says "ESIC provides 70% of average daily wages in cash during medical leave, upto 91 days in two consecutive benefit periods." If the system cannot compute the benefit interaction, your team will.
Test it by running your last real cycle, not the demo data.
Does attendance survive your actual sites?
Most attendance modules work for one office with one shift. In a 1,200-person operation across 4 sites, that assumption breaks in week 1. The test is a second location. A night shift. A device that drops offline for an afternoon.
- Multiple shift patterns, including rotating ones
- Location-specific holiday calendars
- Offline capture that tallies when the link returns
- Overtime rules that match your state
- Comp-off against national and festival holidays
Attendance errors do not stay in attendance. They become loss-of-pay figures and then wrong salaries.
Does it hold the records benefits depend on?
Benefit claims are only as good as the data behind them. ESIC says "ESIC provides reasonable Medical Care for self and family from day one of entering into insurable employment", which means dependant and KYC data has to be right from the joining date, not from confirmation.
ESIC also says "ESIC provides monthly payment apportioned among dependants in case of death due to employment injury." A system that treats dependants as an optional field is storing up a problem a family will discover at the worst moment.
Will employees actually use self-service?
Self-service is the feature that decides whether HR's workload falls. If staff cannot apply for leave, see a balance, get a payslip and file a tax form unaided, nothing changes.
Judge it on a phone, not a laptop. And judge it as a factory worker would, not as an HR manager would.
Is it one employee record, or several?
This is the core question behind every feature list. If leave, attendance, payroll and performance each hold their own copy of the employee, somebody tallies them every month.
One record means an approved leave reaches payroll with no export. A joiner's PF and ESI details fill from where they were collected. That is the difference between a platform and a set of tools sharing a login.
Can you get a report without asking the vendor?
Ask for 3 reports during the trial. Headcount by location. Leave liability. A statutory filing summary. If any of them needs a support ticket, you are buying a system that cannot answer questions.
Reporting is where an HRMS either earns its keep or quietly becomes a data-entry chore.
Which features can you safely ignore?
Anything built to impress in a demo that you will not switch on in year 1. Social feeds, gamified praise, AI summaries of things you were not reading anyway.
They are not harmful. They are just not why you are buying. A vendor leading with them is answering a question you did not ask.
How should you actually evaluate?
| Step | What it proves |
|---|---|
| Run your last real payroll cycle | The statutory engine works on your data |
| Add a second location and a night shift | Attendance holds up |
| Hand a phone to a non-HR employee | Self-service is usable |
| Ask for 3 reports unaided | You can answer questions yourself |
| Request a parallel-run quote | The vendor is confident in migration |
The employment framework behind the statutory tests is published by the Ministry of Labour and Employment. What the modules cover is on complete HRMS, with payroll and leave management set up to share one employee record.