How we score, in plain language

For each brand we average the ratings of the public reviews we hold from each source, normalised to a 5-point scale. The headline score is the equal-weight average of those per-source averages, so every source counts the same. Review counts are the public reviews we hold and link to; totals reported by the platforms themselves are shown separately for reference. If we don't yet hold rated reviews for a brand, we show the platform-reported score and label it as such.

If we haven't collected any scores for a brand yet, we show “—” and a “Gathering” label rather than a zero. A missing score is not a bad score.

We track public reviews from these platforms and always link back to the original page:

CapterraG2GoogleYelpFacebookInstagramGoogle PlayApp StoreRedditWeb

Individual reviews keep their original rating, date and a link to the source.

Facebook public page comments that evaluate the software are a tracked source when harvested. Facebook has no star rating, so a review carries a rating of 1 (negative) or 5 (positive), and Facebook is included in each brand's equal-weight average like any other source. We never use private groups.

Duplicated reviews are shown once. Reddit posts that only mention a name in passing are filtered out.

An automated harvester collects new public reviews and platform scores on a regular cycle (typically weekly). Each per-source score shows when it was last harvested. New reviews are added; existing ones are never rewritten to change their meaning.

• No brand pays to appear, rank higher or remove reviews.
• We don't write reviews and we never invent them. Empty states stay empty until real reviews arrive.
• Curators may hide content only for spam, personal data or abuse, never because it is negative.

Spotted something wrong or outdated? Check the linked original source first. Corrections are assessed against the published review or platform score; a negative rating alone is never a reason to remove a review.