Reviews
Part of Comparing payroll software with statutory scenarios and service tested under pressure
How to review payroll software with a safe test pack and a controlled scoring model
A reproducible payroll software review method for England covering scope, official recognition, realistic tests, security, service, price and disclosure.
A useful review begins with an employer context, not a star rating. This method is designed for desk research and controlled trials of payroll products used by employers in England. It does not claim hands-on testing until a named reviewer has completed and recorded that work.
State the review population
Record employee count, weekly or monthly pay, variable hours, statutory payments, directors, pension arrangement, benefits, multiple PAYE schemes and required integrations. Exclude any product that does not support a mandatory scenario. A simple micro-employer and a multi-site bureau should not receive the same weighting.
HMRC's software-selection guidance lists functions that may differ, including payslips, pension deductions, different pay periods and some reports. Turn those differences into pass or fail requirements before looking at brands.
Verify official recognition precisely
Check the current product name on HMRC's recognised payroll software list. Record the date, supplier and exact edition. Recognition means the listed product can report PAYE information online. HMRC explicitly says it cannot recommend one product over another and is not responsible for purchased software problems.
Recognition is therefore an eligibility check, not a quality score. It does not by itself prove pension exchange, service performance, usability, security or suitability for a particular employer.
Build a safe test pack
Use synthetic workers and a non-production environment. Include starters and leavers, irregular pay, statutory pay, pension assessment, corrections, rejected submissions, year end, permissions and data export. Write expected results from current official guidance, then have a payroll specialist review the pack.
Score outcome accuracy, evidence shown to the operator and recovery from error. A fast calculation that cannot be explained or corrected should not outrank a slower, controlled workflow.
Examine service and resilience
Test support with the same payroll-critical questions and record channel, authentication, first meaningful response and resolution. Request service definitions, planned maintenance, incident communications, backup responsibilities, recovery evidence and exit assistance. Supplier statements remain unverified until tested or supported by contractual evidence.
Review access control, audit logs, encryption, subprocessors, retention and breach handling with security and privacy owners. The ICO's worker-information guidance is a relevant official starting point because payroll contains personal information, but a product assessment still needs its own data mapping and risk analysis.
Compare total cost
Normalise the quote to one defined year and workforce. Include base subscription, employee or run charges, pension functions, payments, implementation, migration, training, support tiers, integrations, storage, price changes and exit. Mark promotional pricing separately and keep VAT treatment explicit.
Control the scoring model
Use a small number of defined criteria and publish what each score means. A five-point accuracy scale, for example, needs observable anchors rather than a reviewer's general impression. Keep mandatory pass or fail controls outside the weighted total so that attractive usability cannot cancel a failed statutory journey.
Run the calculation with alternative reasonable weights. If the winner changes whenever support or migration receives a modest adjustment, present a shortlist by employer type instead of declaring one overall leader. Include not tested and insufficient evidence as valid results; treating missing evidence as average rewards opacity.
Have a second person repeat the decisive cases without seeing the first score. Investigate disagreement and retain both records. Product defects, tester mistakes and ambiguous instructions require different conclusions.
Publish the evidence record
Name the reviewer, competence, test date, version, device, browser, scenario pack and assistance received. Disclose free access, affiliate arrangements, advertising and supplier fact checks. The CMA's reviews and endorsements collection gives businesses current official material on fair review practices.
Publish criterion weights before the final score. Show raw findings and blocking failures, not just a composite number. Give suppliers a chance to correct factual descriptions without allowing them to approve the verdict. Recheck recognition, features and prices on publication day, and date every conclusion.