Ranking method
Every ranking answers one question for a stated reader, on a rubric we publish before we gather the data. We record each criterion with a definition, a unit, a data source and a weight with the reason for it. We date every value and cite every source.
The rubric
Every comparison has its own rubric, published on that comparison's page before we gather the data. A rubric lists each criterion with:
- a definition and a unit;
- the data source we will use;
- a weight, and the reason for that weight.
The criteria and the weights change from comparison to comparison, because the question changes. Read the rubric of a comparison on its own page. The first one is Invoicing software review 2026.
Where we choose not to score a criterion — ease of use in a software comparison, for example — we say so on that comparison and show the underlying data, so you can judge for yourself.
Data sources
We prefer, in order: our own documented testing; primary data (the product's own pages, regulators, official statistics); open licensed datasets; reputable independent testing; and, last, user ratings with their sample size and date.
For a price or plan we use the vendor's own pricing page as the source and record the retrieval date. Prices and plans change — always check the vendor's page before you decide.
How the order is decided
Scores are combined with the published weights. Commissions never change the order: the method is fixed before any commercial relationship, and if a relationship ever exists it is disclosed on the piece and on Disclosures. Where a comparison has no single overall order (for example when the best tool depends on your situation), we publish a per-situation verdict instead of forcing a ranked list.