Methodology
How we test software
Every score on this site comes from the same process. Here is exactly what a review has to go through, and how to read the state it carries.
Three review states
Every product on the site carries one of three states, and you always see which one you are reading. As of July 2026 every review is research based, with hands-on testing rolling out; when a product is tested, its review is updated, re-dated and moved to the tested state.
Hands-on Tested. We create a real account, pay for or use the actual plan, and run the core workflows a business runs. The tested stamp shows the month we ran it and the plan or version, so you can judge how current it is.
Research based. Assessed from documentation, pricing and aggregated user feedback, with sources shown. It carries a score, but we have not run it hands-on yet, and we say so.
Listed, not reviewed. A known option we track so the category picture is complete. No score until we test or research it properly.
What a hands-on test involves
For accounting software, that means creating and sending a real invoice, importing a bank statement and reconciling a month of it, generating a tax return and checking the figures, and walking any local e-invoicing setup. Other categories have their own core workflows, listed in each review. We test on the plan most SMEs would actually buy, not a stripped demo.
How we score
We grade on a fixed set of criteria per category, each out of 10, then weight them toward what a small business feels day to day: ease of use and local compliance count for more than raw feature depth. The overall is not a simple average; the weighting is described on each review. Scores can move when software changes, and we re-date the review when they do.
Local compliance
We check the things US and UK reviews skip: GST and SST, IRAS InvoiceNow and LHDN MyInvois, PayNow and DuitNow, CPF and EPF payroll. A product that cannot handle the local rule an SG or MY business needs is marked down for it, however polished it is elsewhere.
Re-testing
SaaS changes fast. Each hands-on badge flags for a re-test after twelve months, and we update sooner when a product ships a change that affects the verdict. If we get something wrong, we correct it and note the change; see our editorial policy.