A payroll quality checklist covering rules, inputs, approvals, submissions and recovery

Quality means the defined payroll workflow produces correct, authorised, traceable and recoverable outcomes. It is not demonstrated by a clean interface or one successful happy-path test. Apply this checklist to a named product version, tax year and customer scenario.

Scope and rules

• Record supported employers, employees, pay frequencies and exclusions.
• Inventory applicable calculations, fields and submission types.
• Link every rule to a current official source and effective date.
• Assign change monitoring and regression-test owners.
• Prevent use of unsupported tax-year configurations.

HMRC's developer support collection links technical specifications and test data. Use the relevant material with wider product and workflow tests.

Input and calculation

• Validate mandatory data, effective dates and duplicates.
• Test ordinary, boundary, invalid and combined cases.
• Cover starters, leavers, variable pay and relevant statutory items.
• Preserve overrides with reason and approval.
• Reconcile totals and material variances before release.

Use synthetic records for routine testing. Protect any controlled production data used for investigation.

Access and approval

• Apply least privilege and strong administrator authentication.
• Separate customers and PAYE schemes in bureau use.
• Record important configuration, import and approval actions.
• Remove leaver and temporary implementation access promptly.
• Require explicit authority for submissions and payment outputs.

Reporting and correction

• Test FPS and EPS generation in the supported context.
• Preserve acknowledgements, rejections and retry state.
• Prevent ambiguous duplicate submissions.
• Provide a reviewed correction workflow by error type and period.
• Reconcile payroll records with expected HMRC outcomes.

HMRC's error-correction guidance shows that the action differs for pay, dates, employee details, National Insurance and EPS errors. Do not build one "undo" control for every situation.

Pensions and connected systems

• Test the named scheme, tax-relief method and provider format.
• Exercise assessment, worker requests and communications in scope.
• Confirm provider acceptance and contribution reconciliation.
• Validate accounting, time and payment mappings.
• Make failed interfaces visible to an accountable operator.

Resilience and support

• Back up critical data and configuration.
• Restore into a controlled environment and verify completeness.
• Rehearse dependency loss, interrupted processing and account recovery.
• Match support and escalation to customer payroll schedules.
• Communicate incidents and workarounds through approved routes.

The NCSC's secure SaaS guidance covers authentication, logging, integration control, incidents and backups.

Log each result as passed, failed, unknown or not applicable with evidence and owner. A release can proceed only when authorised reviewers understand unresolved risk and stop conditions. Re-run decisive checks after tax-year, integration or architecture change.

Migration and cutover

• Reconcile employee identities and year-to-date values.
• Preserve payroll IDs or handle changes through current guidance.
• Run representative old and new payrolls in parallel.
• Investigate every difference without a blanket tolerance.
• Define authoritative system, rollback trigger and cutover approval.

Employee-facing quality

• Deliver payslips and documents to the correct authenticated user.
• Test keyboard, screen-reader and mobile access with relevant users.
• Keep confidential values out of ordinary notification messages.
• Provide an approved support and alternative-delivery route.
• Verify leaver access and retention against the employer process.

Evidence and review

• Link requirements, tests, defects and release in one record.
• Identify who reviewed every critical pass.
• Keep an unresolved-issue log with customer effect and expiry.
• Compare post-launch outcomes with test assumptions.
• Feed repeated corrections back to source controls and regression cases.

A quality dashboard should separate files created, files sent, receiving-system acceptance and final reconciliation. Measure correction time as well as technical recovery. Review failed and skipped checks after each significant incident, and withdraw unsupported scope instead of training users to work around a known defect.