Loading page
All-in-one HR platform · v2.0
14-day free trial · No credit cardhello@HrMagix.com
All-in-one HR platform · v2.0
14-day free trial · No credit cardhello@HrMagix.com
The hard part of a first payroll run is almost never the calculation. It is establishing what is true, which salary structure a person is actually on, which identifiers are correct, and what the previous system has already reported to the government on your behalf.
This guide runs in the order the work runs. If you are switching at the start of a financial year, chapters four and five get considerably shorter; everyone else should read them carefully.
01
There are two sensible moments. The start of a financial year is by far the cleaner: no year-to-date figures have to travel, and the annual certificate at the end of the year is produced entirely by one system.
Mid-year is the common case, because the decision to change is rarely taken in March. It is entirely workable, but it carries one hard requirement that people underestimate: every rupee already paid and every rupee of tax already deducted in the current financial year has to come across, because Form 16 must reconcile across the whole year regardless of how many systems produced it.
Watch for itDo not switch in the same month as a salary revision cycle. You will be trying to prove two things at once, and when the numbers disagree you will not know which change caused it.
02
Everything downstream references the employee record, so it goes in first. This is also the point at which you discover what your spreadsheets have disagreed about for two years.
Two problems surface almost every time, and neither is caused by the migration. Duplicate records for the same person, usually created when somebody was rehired or moved between entities. And salary structures that were described one way in an offer letter and calculated another way in practice.
03
Statutory deductions are derived from the salary structure and the registrations you hold, not entered by hand. That means the registrations have to exist in the system before a run can compute anything correctly.
For a multi-state employer this is the step that takes longest, because professional tax is a state subject: you hold a separate registration in each state where you employ people, and the schedules differ between them.
04
If you are switching mid-year, this is the chapter that determines whether your annual certificates are correct. For each employee you need, for the financial year so far: gross paid, each statutory deduction made, and tax deducted and deposited.
The reason is narrow and unavoidable. Tax under Section 192 is projected across the whole financial year, so the new system needs to know what has already been deducted in order to project the rest correctly. Get this wrong and the error does not show up in the first month, it shows up in February, when the projection corrects itself violently in one deduction.
Watch for itReconcile the year-to-date figures against the Form 24Q returns already filed, not against your internal spreadsheet. The return is what the tax authority has been told; that is the number that has to match.
05
Process the same month in both the outgoing system and the new one, and compare the results employee by employee. Do not compare totals: two offsetting errors produce a matching total, and you will have proved nothing.
Differences cluster in a small number of predictable places, which is what makes this exercise finite rather than open-ended.
06
Once the parallel month agrees line by line, the first live run should be unremarkable. Two habits are worth forming on that first close, because they are much harder to introduce later.
Lock the period once it is processed, and make corrections as identified adjustments in a later run rather than by editing a closed month. The closed month has already been filed against; editing it without a record makes the filing and the ledger disagree. And keep the payslip as issued rather than regenerating it, so a reprint two years from now is the original document.
Every guide ends here rather than in a summary. If a line below is not true yet, that is the next thing to do.
The start of a financial year is the cleaner moment, because no year-to-date figures have to travel and the annual certificate is produced entirely by one system. Mid-year is workable, but every rupee paid and every rupee of tax deducted so far that year has to come across.
The guide advises against it. You would be trying to prove two things at once, and when the numbers disagree you will not know which change caused it.
The employee master: full name as on statutory records, date of joining, department, location and reporting line, salary structure by grade, statutory identifiers (UAN, ESIC number where applicable, PAN, bank account) and entity where there is more than one.
The EPF establishment code and your wage-ceiling choice, the ESIC code for each covered establishment, a professional tax registration for every state you employ in, labour welfare fund registration where the state operates one, and your TAN for Section 192 TDS.
Tax under Section 192 is projected across the whole financial year, so the new system needs to know what has already been deducted. If this is wrong, the error tends to surface in February, when the projection corrects itself in one deduction.
Against the Form 24Q returns already filed, not your internal spreadsheet. The return is what the tax authority has been told, so that is the number that has to match.
Employee by employee and head by head, not in total, because two offsetting errors produce a matching total. Differences usually cluster around PF rounding, allowances treated differently for ESI, mid-year ESI eligibility changes, loss-of-pay days and arrears.
The guide recommends locking the period once processed and making corrections as identified adjustments in a later run. Editing a closed month that has already been filed against makes the filing and the ledger disagree.
The checklist above is generic by necessity. Bring your own structures, states and rota and we will tell you which lines are already true and which are not.
14-day free trial · Full access to all 12 modules · Zero implementation fee · Cancel anytime