Data quality dashboard: what to show, to whom and how to build it step by step
Many data quality dashboards are built with enthusiasm and abandoned within a month. The reason is usually the same: too many indicators, none with an owner and none with an agreed threshold. Here is how to build one that actually gets used.
Why many data quality dashboards go unused
- Too many indicators: a hundred traffic lights don’t help anyone decide.
- No agreed threshold: a red light nobody defined obliges nobody to act.
- No owner: if a KPI has no owner, nobody fixes it.
- No history: without a trend you can’t tell whether things are improving.
Three views for three audiences
1. Management
A single screen: the overall quality index by data domain (customers, products, finance…), its trend over the last 6 to 12 months and the three issues with the biggest business impact. Nothing technical.
2. Data owners
Each data owner or data steward sees their KPIs against their threshold, how they evolve and the open issues in their domain, with their age.
3. Technical team
Which rules fail, how many records they affect, which system or process they come from and when the last check ran. This is the view for fixing.
Which indicators to include
- Five or six KPIs on critical data, one per relevant dimension. Here are 15 examples with formulas.
- The overall index for each domain: the weighted average of its KPIs.
- Open issues and average resolution time.
How to build it in six steps
- Pick the critical data: the few fields that decisions or regulatory reports depend on.
- Define rules and thresholds with whoever uses the data. What level is acceptable.
- Automate measurement: scheduled queries or the quality tool you already have, storing each result with its date in a history table.
- Design the three views on top of that history table.
- Assign an owner to each KPI and each domain.
- Review it monthly in the data committee, with decisions: what gets fixed, by whom and by when.
Which tool to use
The one you already have. Power BI, Looker Studio or Tableau all work, and even a spreadsheet to start. Quality tools such as Great Expectations or Soda, or the quality modules of data governance platforms, produce the results; the dashboard shows them. Comparison in data quality tools in 2026.
Common mistakes
- Starting with the tool instead of the critical data and thresholds.
- Showing everyone the same thing: management doesn’t need technical rules.
- Measuring without deciding: if the committee doesn’t assign fixes, the dashboard is decoration.
Frequently asked questions
How often should a data quality dashboard be refreshed?
The technical view daily or with each load. The management view only needs a monthly review.
Do I need a paid tool?
Not to start. Scheduled queries and the reporting tool you already use are enough for a first version.
What is the overall quality index?
The weighted average of a data domain’s KPIs, giving more weight to the dimensions that matter most for that use. It’s the figure reported to management.
Informational content.
How good is your data?
Calculate your data quality score by dimension for free, no sign-up, and see what to fix first.