Interpreting Quality Scores¶

DO Calculating quality scores

A quality score is a number from 0 to 100, but reading it well takes context: what the data is used for, which dimensions pull it down, and which containers feed the aggregate. This page shows where scores appear and how they support your decisions, and explains how to read them as a measure of fitness for purpose. For a datastore score that looks lower than expected, see the walkthrough in Examples.

Where to See the Score¶

Datastore-level quality scores are offered for source datastores, not enrichment datastores. Enrichment datastores can still have checks, but they do not get a datastore quality score and are left out of estate-level quality insights.

The quality score is visible across the platform without needing to navigate to a dedicated page:

• Datastores listing page: Each source datastore row displays its current quality score.
• Datastore detail page: The score appears in the Totals panel at the top of the Overview tab.
• Container detail page: Each table or file displays its own container-level score.

How Scores Influence Your Datastore¶

The quality score gives data stewards a shared view of measured data health. Compare totals across assets only when their decay periods and dimension weights are comparable, and review the underlying dimensions and settings alongside each score. You can use it for:

• Prioritizing investigation: Compare scores, then drill down to the dimensions, containers, and fields contributing to a lower result.
• Tracking improvement: After addressing a data issue and running the relevant operation again, review how the newly recorded score changes the trend.
• Communicating data health: Use scores and dimension details to explain measured quality and changes to stakeholders.
• Reviewing recent results: Scores update when relevant operations or quality events change their inputs, so the latest score reflects the most recently recorded signals.

How to Interpret and Use Quality Scores¶

Quality scores are dynamic measures of confidence that reflect the intrinsic quality of your data. It's important to recognize that different types of data will have varying levels of inherent quality. To illustrate this point, let's consider a standard mailing address in the USA. A typical schema representing a mailing address includes fields such as:

• Addressee
• Street
• Street 2
• City
• State
• Postal Code

The "State" field, which is naturally constrained by a limited set of known values, will inherently have a higher level of quality compared to the "Street 2" field. "Street 2" typically holds free-form text ranging from secondary unit designations to "care of" instructions and may often be left blank. In contrast, "State" is a required field for any valid mailing address.

Consider the level of confidence you would have in making business decisions based on the values held in the "State" field versus the "Street 2" field. This thought exercise demonstrates how the Qualytics Quality Score (with default configuration) should be interpreted.

While there are steps you can take to improve the quality score of the "Street 2" field, it would be unrealistic to expect it to meet the same standards as the "State" field. Instead, your efforts should focus on the change in measured quality score over time, with the goal of raising scores to an acceptable level of quality that meets your specific business needs.

To further explore how to respond to Quality Scores, let's consider the business requirements for capturing "Street 2" and its downstream use:

• If the primary use case for this address is to support credit card payment processing, where "Street 2" is rarely, if ever, considered, there may be no business need to focus on improving the quality of this field over time. In this case, you can reduce the impact of this field on the overall measured quality of the Address by applying a Tag with a negative weight modifier.

• On the other hand, if the primary use case for this address is to reliably ship a physical product to an intended recipient, ensuring a higher level of quality for the "Street 2" field becomes necessary. In this scenario, you may take actions such as defining additional data quality checks for the field, increasing the frequency of profiling and scanning, establishing a completeness goal, and working with upstream systems to enforce it over time.

See Also¶