A data quality score you can actually explain to your team.

“Our data quality is 82%” means nothing if nobody can say why. Sohovi's scoring engine is built so every number — column score, dataset score, or catalog score — traces back to a specific rule, a specific formula, and a specific set of failing rows.

10 ISO-standard dimensions

Every dataset is evaluated across ten dimensions: completeness, accuracy, consistency, validity, uniqueness, integrity, timeliness, currency, conformity, and precision. Each dimension has a plain formula — for example, completeness is the percentage of non-null values in a column, and validity is the percentage of values that match a defined rule (a regex, a range, a lookup list, or a cross-column check). There's no proprietary weighting hidden behind the number.

Column, dataset, and catalog level

Scores roll up from individual rules to a column score, from columns to a dataset score, and — on the Business plan — from datasets to a catalog-level score across an entire business unit. A score under 60 is flagged critical, 60–80 is a warning, 80–95 is good, and 95+ is excellent, so anyone on your team can read the color and understand the severity instantly.

Drill into any failure

Click into any dimension and see the exact rule that was applied, how many rows passed versus failed, and a preview of the failing records themselves. This transparency panel is what turns a quality score from a vanity metric into something you can act on — Sohovi tells you which rows to fix, not just that something's wrong.

Track quality over time

Every run is saved to your score history, so you can chart whether quality is improving or degrading across weeks or months, catch schema drift the moment a new or renamed column shows up, and set alert thresholds that notify you before a quality regression reaches production. Rules are suggested automatically based on your column profile — see how that works, or check what's included per plan on our pricing page.