Datastore Quality Score FAQ¶

Common questions about how scores are calculated, their settings, why they change, AI explanations, and access.

Scores¶

What is a Data Quality Score?¶

A Data Quality Score is a measure from 0 to 100 that reflects the health of your data. Higher scores indicate stronger measured quality. Scores are calculated at the field, container, and datastore levels, and recorded over time so you can track how they change. See How Datastore Quality Scores Work.

How is the datastore score calculated?¶

The datastore score is the weighted average of its container scores. Only containers that have a score, meaning they have been both profiled and scanned, are included. The tag weight modifiers on each container decide how much it counts. See Datastore-Level Scoring.

How is the container score calculated?¶

Each container score uses a multiplicative baseline model that starts from a baseline of 70. Each of the 8 quality dimensions then moves the score up or down in proportion to its configured weight, and the result is kept between 0 and 100. See Container-Level Scoring.

Why does the model start from a baseline of 70?¶

The baseline of 70 is the fixed point the model starts from before any dimension adjusts it. It is not the score a new container should expect: dimensions that have not been measured yet can count against the score, so a freshly profiled container with few checks can score well below 70 even with no anomalies. Each measured dimension then pulls the score down when issues exist, or pushes it slightly up when the dimension is exceptionally clean. The baseline is not configurable.

What are the 8 quality dimensions?¶

Completeness, Coverage, Conformity, Consistency, Precision, Timeliness, Volumetrics, and Accuracy. Each measures a different aspect of data quality and is scored independently from 0 to 100. See The 8 Quality Dimensions.

What do a score of 0 and a score of 100 mean?¶

A score of 0 indicates severe data quality issues, such as a container where every check fails or where dimensions like Accuracy or Consistency have major problems. A score of 100 indicates the best measured quality across all counted dimensions: no anomalies, all checks passing, and complete data with consistent types and volumes.

Why doesn't my datastore have a quality score?¶

A datastore gets a score only once at least one of its containers has a score, and a container needs both a completed Profile and a completed Scan. Run Sync, then Profile, then Scan on the tables you want to monitor. Enrichment datastores never get a datastore quality score. See Troubleshooting.

Why is my datastore score different from the average of my container scores?¶

The datastore score is a weighted average, not a simple one. Containers with higher tag weight modifiers contribute more. Containers without a score, because they have not been both profiled and scanned, are left out of the calculation.

Does syncing affect the quality score?¶

Not directly. Sync discovers containers and fields but does not produce scores. Containers discovered by Sync show no score until they are profiled and scanned, and only then join the datastore score.

Settings¶

Do settings on the datastore affect my containers?¶

It depends. Datastore settings act as the default for any container without settings of its own, so changing the datastore's decay period or dimension weights affects those containers. A container with its own settings is unaffected, and its settings do not change the datastore or other containers. See Settings Inheritance and Overrides.

Can I configure settings on a field?¶

No. Fields do not have settings of their own. A field's score is calculated with the settings of its container, so to change them you edit the container's settings. See Configure Quality Score Settings.

Can different datastores have different settings?¶

Yes. Settings are configured per datastore, so each datastore can have its own decay period and dimension weights.

What is the decay period?¶

The decay period controls how far back in time Qualytics looks when calculating scores. Only anomalies, profiles, and scans within the window count. The default is 180 days, and you can set any value from 7 to 180 days in Configure Quality Score Settings or through the API.

What happens when anomalies age out of the decay window?¶

They are no longer counted. With a decay period of 180 days, an anomaly detected 7 months ago no longer affects the score, even if it was never resolved. The score improves as old issues age out.

What are dimension weights?¶

Each of the 8 dimensions has a weight from 0 to 2.0, in steps of 0.1, that controls its impact on the container score. The default is 1.0, and a null value in the API also resolves to 1.0. See Dimension Weights.

What happens if I set a weight to 0?¶

The dimension is effectively disabled. It turns grey in the settings dialog, is treated as fully healthy, and cannot pull the score down.

Can I set a weight higher than 1?¶

Yes, up to 2.0. A weight of 2.0 doubles the dimension's impact compared to the default, making the total score more sensitive to it.

Does changing the settings recalculate existing scores?¶

No. Already-recorded scores keep the settings that were in effect when they were measured. The new settings apply from the next score event, such as the next Scan or Profile. This is why the weight badge on each dimension row shows the weight at the time of measurement, not the current setting.

Score Changes¶

When are scores recalculated?¶

Quality scores are recalculated automatically when:

• A Scan operation completes (anomalies detected or clean scan).
• A partition-level scan finds new anomalies.
• A scan is deleted.
• A Profile operation completes (new field statistics).
• An anomaly status changes (acknowledged, resolved, etc.).
• An anomaly is deleted.

Are recalculations immediate?¶

They happen automatically after qualifying events. When many events occur in quick succession, such as resolving anomalies in bulk, they are batched to avoid redundant calculations.

Can I manually trigger a recalculation?¶

No. Scores are always recalculated automatically, based on the events above.

Why did my score change without running a new scan?¶

Several events recalculate the score without a new scan:

• Resolving or acknowledging anomalies changes the anomaly count.
• Deleting an anomaly removes it from the calculation.
• A partition-level scan that finds new anomalies produces a new score.
• Anomalies aging out of the decay window improve the score.

Why isn't my score improving after fixing issues?¶

Qualytics does not know the data is fixed until a Scan confirms it, so run a Scan after fixing the source data. Another dimension, or a dimension with a high weight, may also still be holding the score down. See The Score Does Not Improve After Fixing the Data.

Why does the number of checks next to a dimension not match its score?¶

The count pill on a dimension row is taken when the Quality Score panel opens: it lists the dimension's checks that have an open anomaly inside the decay period right now. The score is a snapshot from the last scoring run. Between the two, anomalies can be opened, acknowledged, archived, or age out, so a dimension can score well and still list checks, or show no pill while its score has not caught up yet. Only Conformity, Precision, Accuracy, Timeliness, and Volumetrics show the pill, because their scores are driven by check anomalies. See Checks with open anomalies.

How do tags affect quality scores?¶

A tag can carry a weight modifier from -10 to 10. The modifiers of the tags on each container decide how much that container counts in the datastore score, so containers with higher tag weights have more impact. See Tag Weight Modifiers.

What happens when I add or remove a tag with a weight modifier?¶

Container weights are recalculated when tags are added to or removed from the datastore, and when you use Move to, Merge, or Remove on a tag's page (Move a Tag to Another Tag, Merge Tags, Remove a Tag from Assets). Both update the datastore score right away. A tag assigned directly to a container does not change its weight until one of those happens, so until then the datastore score keeps the previous weights.

AI Explanations¶

Why does one datastore show a sparkle icon on its Quality Score and another does not?¶

The sparkle appears on the Quality Score chip only when AgentQ is configured for the workspace. The signal is workspace-wide, so when one datastore shows it and another does not, the difference is whether the score has finished being enriched with AI content. A datastore whose score was just measured may briefly show no sparkle until the AI content lands.

What if AgentQ is not configured yet?¶

The Quality Score panel still opens and shows the score, the dimension rows with their static definitions, the sparklines, the chevron stacks, and the weight badges. The AI summary, the per-dimension AI explanation lines, and the AI-written advice are absent. Users with permission to manage integrations see a banner at the bottom of the panel that links to the AgentQ setup.

Why does the Recommendations section not appear for one of my datastores?¶

The section is hidden when no dimension scores below 50. A dimension must score strictly below 50 to get a recommendation, so a score of exactly 50 does not qualify.

Do container and field Quality Scores get AI explanations too?¶

Yes. The same panel opens from every Quality Score chip, whether it belongs to a datastore, a container, or a field. The AI reasons about the signals of that level: the worst-scoring containers on a datastore, the top failing checks on a container, or the field's profile on a field. See View AI Explanations and Recommendations.

Access and API¶

Who can view quality scores?¶

Any user with at least the Member user role and the Reporter team permission on the datastore. See Permissions.

Who can change quality score settings?¶

Users with the Member user role and the Editor team permission on the datastore. See Permissions.

Is there an API to update quality score settings?¶

Yes. Use PUT /api/datastores/{id}/score-settings for a datastore, or PUT /api/containers/{id}/score-settings to override the settings of a single container. Each call replaces the full settings object. See Update Quality Score Settings for examples.

Can I retrieve historical quality scores via the API?¶

Yes, for datastores, containers, and fields: GET /api/datastores/{id}/quality-scores, GET /api/containers/{id}/quality-scores, and GET /api/field-profiles/{id}/quality-scores. Each returns the 10 most recent scores, one per day by default, or one per measurement with granularity=measure. There is no pagination or date range parameter. See Get Historical Quality Scores.