Indian payroll is not difficult arithmetic. It is a lot of small rules, each with its own base, its own ceiling and its own authority — and the failure mode is a small mistake repeated twelve times before anyone notices. Attendance, leave and payroll on one record, with the four statutory contributions computed properly.
The mistake that causes the most rework is treating these as four percentages. They are four different shapes of rule, and software that models them as one thing with four rates will be wrong in a way nobody spots for months.
A percentage of a defined wage base, with the statutory ceiling above which coverage stops being mandatory, the employer share split between pension and provident components, and administrative charges on top. Computed with its ceiling rather than as a flat rate.
Separate employee and employer rates applied below a wage threshold. The behaviour at the threshold is where most systems go wrong, because an employee crossing it mid-period is governed by the contribution period rather than by the month.
A state levy, not a central one. Slabs, periodicity and caps all differ by state and several states do not levy it at all. Employ people in three states and you have three sets of rules — a single PT field is not handling professional tax.
Also state-specific, also with its own periodicity — half-yearly in some states rather than monthly — and small enough per employee that it is the one most often forgotten until an inspection.
Rates, ceilings and slabs are changed by notification and several vary by state, so treat any figure you read anywhere — including in a vendor's marketing — as a pointer to check against the current rule rather than a constant.
PF, ESI, PT and LWF computed with their ceilings, thresholds and slabs rather than as four percentages — because they are four different shapes of rule.
Annual liability projected and spread across remaining months, reprojected when a declaration changes, so nothing lands on the employee in March.
Produced from the same records that computed the deduction, rather than assembled separately afterwards where the two can drift apart.
The days that drive the pay are the days the system already recorded. No monthly reconciliation between an attendance tool and a payroll tool.
Carrying the business's own logo — something an employee can hand to a bank or a landlord, not a spreadsheet row.
Pay and bank details sit behind their own routes rather than behind a flag on a general permission — a separation that holds when someone edits a setting.
Payroll bought on its own leaves somebody reconciling it against everything else every month. On one record engine that reconciliation does not exist to be done.
Written up at greater length in PF, ESI, PT and LWF: Indian payroll without the March panic, including five test cases worth running against any payroll system before you buy it.
Rise Up is our time tracker. It is a separate product, not a module — but the hours it records sync into Orbis, so the figure somebody is paid on is the one that was measured rather than one re-typed from a timesheet at the end of the month.
Linux agents are genuinely uncommon in this category; most trackers cover macOS and Windows and leave Linux users filling in a form.
Yes, and with their ceilings, thresholds and slabs rather than as flat rates. That distinction matters more than it sounds: the four are different shapes of rule, and a system that models them as one thing with four percentages will be wrong in ways nobody notices for months.
And an employee in a second state. Those two cases between them expose more about a payroll system than an hour of slides, and we would rather you tested it that way.