Why payroll software projects fail in Indian MSMEs, and how to avoid the usual patterns

Payroll software projects in small businesses rarely fail because the software cannot calculate. They fail because inputs are wrong (employee masters, attendance, salary structures), because rules were set up once and never updated (labour code wages, the EPF ceiling, state PT), because a mid-year switch broke year-to-date figures, or because no one owns payroll after go-live. The patterns below are generic, drawn from how payroll works, not from any named company. Each has a simple control.

This guide links to every guide in the failure patterns, migration and sources section.

Pattern 1: Dirty employee masters

What happens. Missing UANs, wrong ESI numbers, PAN not linked, outdated bank accounts, wrong joining dates. Challans are rejected, TDS credits go to the wrong PAN, salaries bounce.

Control. Clean masters before migration: reconcile UANs with the EPFO portal, ESI numbers with ESIC, PAN with the e-filing portal, bank details with a ₹1 test or the bank's validation. Lock bank account changes behind approval. See the migration guide.

Pattern 2: Attendance never reaches payroll on time

What happens. Biometric data is exported late, missed punches are fixed by email, overtime is added manually. Payroll is run on estimates and corrected next month, producing arrears and wrong PF and ESI.

Control. Integrate attendance directly, set a cut-off date, require regularisation in self-service before cut-off, and lock attendance before the run. See integrations with Tally, banks and biometrics.

Pattern 3: Salary structures that fail the 50% wage rule

What happens. Old structures with low basic and large allowances carry over. PF, gratuity and ESI are under-calculated under the labour codes, creating arrears and liability.

Control. Run the 50% test on every structure, show the add-back line in payroll, and restructure with advice. See labour codes and payroll.

Pattern 4: Rule updates missed

What happens. The EPF ceiling rose to ₹25,000 from 17 September 2026, a state revised PT or minimum wages, the Income-tax Rules, 2026 renumbered forms. A system configured once keeps applying old values.

Control. Choose software where statutory updates are included and dated in release notes; assign someone to read EPFO, ESIC, state and income-tax notices monthly; keep the official sources list at hand.

Pattern 5: Mid-year switch breaks TDS and PF history

What happens. A new system starts in October without April–September salary and TDS. Projections restart, employees are over- or under-deducted, Form 138 for the year mixes two sources and Form 130 is wrong.

Control. Migrate year-to-date figures per employee, run a parallel month, and prefer a quarter start for go-live. See migration.

Pattern 6: Bank file edited by hand

What happens. Someone edits the salary upload file to fix a last-minute change. Totals no longer match the payroll register; a payment goes to the wrong account; fraud is possible.

Control. Generate bank files from the approved payroll only, use maker-checker approval, reconcile bank debits to the register each month.

Pattern 7: Full and final settlements too slow

What happens. Settlements take weeks because approvals, asset return and leave balances are chased by email, though wages due on exit must be paid within two working days under the Code on Wages.

Control. An off-cycle settlement workflow with a two-day target, leave balances visible in real time, gratuity and bonus computed separately. See gratuity, bonus and leave encashment.

Pattern 8: Wrong tier, wrong type

What happens. A business buys payroll-only software and runs shifts in spreadsheets, or buys a full HRMS and uses only payroll. Either the team does manual work or pays for modules nobody uses.

Control. Choose by fit with the best-fit method and compare types on the compare page.

Pattern 9: Contract silent on errors and exit

What happens. A statutory error causes interest or damages, and the contract says nothing about vendor responsibility. At exit, historical registers and tax data cannot be exported.

Control. Negotiate update SLAs, error handling, data export and retention. See contract, SLA and exit terms.

Pattern 10: Employee data overexposed

What happens. Managers see everyone's salary, payslips are emailed as unprotected files, ex-employees keep app access, vendor staff have broad access.

Control. Role-based access, encrypted delivery, access removal on exit, and a data processing agreement. See employee data under DPDP.

A monthly control routine

When	Check
Before cut-off	New joiners and leavers entered; attendance locked; salary changes approved
After the run	Variance report against last month by employee; exceptions explained
Before bank upload	Register totals = bank file totals; maker-checker approval
By the 15th	PF ECR and ESI contributions filed and paid
Each quarter	TDS projections reviewed; Form 138 data validated
Each year	Bonus, gratuity provisions, Form 130 issued by 15 June

When to call in help

If arrears from labour code wage corrections build up, or if PF, ESI or TDS notices arrive, involve a payroll consultant, labour law adviser or CA early. Fixing filings quickly is cheaper than defending them later.

Frequently asked questions

Is the software vendor responsible if payroll is wrong?

Legally the employer is responsible. The contract can allocate costs of vendor-caused errors, so negotiate it.

How long does it take to stabilise new payroll software?

Plan for a parallel month and two or three live months before the routine settles.

What is the single most useful control?

A monthly variance report by employee, reviewed before bank upload. It catches most input errors.

Are these patterns from real companies?

They are generic patterns based on how payroll processes work, not accounts of any named business.

Sources

EPFO — PF filings, interest and damages
ESIC — ESI filings
Income Tax Department — TDS obligations and forms
Ministry of Labour and Employment — labour codes and rules