Data Quality KPIs Every Leadership Team Should Track
Published 3 August 2026
Why most data quality scorecards fail
They either measure 40 metrics across 200 data sets (nobody reads them) or one metric across one data set (no signal). The middle path: 7 KPIs across the data sets that drive the business.
The seven KPIs
1. Accuracy
Percentage of records correct against a trusted source or business rule. Example: postcode validates against PAF, VAT number matches Companies House.
2. Completeness
Percentage of mandatory fields populated. Example: 98% of contacts have a valid email.
3. Timeliness
Percentage of records updated within the expected refresh window. Example: 95% of stock balances refreshed within 4 hours of the warehouse movement.
4. Duplication
Percentage of records identified as duplicates by your matching rules. Example: 2.1% duplicate customers (down from 7% last quarter).
5. Lineage coverage
Percentage of in-scope data sets with documented lineage from source to consumption. Underpins audit, AI explainability and impact assessments.
6. Owner coverage
Percentage of in-scope data sets with a named business owner. Below 100% on tier-1 sets is a red flag.
7. Issue resolution time
Median days from a data quality issue being raised to being resolved. Trend matters – a stable median proves the operating model works.
Example leadership dashboard
| KPI | Target | This quarter | Trend |
|---|---|---|---|
| Customer accuracy | ≥ 98% | 97.4% | ↑ from 96.1% |
| Contact completeness | ≥ 95% | 93.8% | ↑ from 91% |
| Pipeline timeliness | ≥ 95% | 99.1% | flat |
| Customer duplication | ≤ 3% | 2.1% | ↓ from 4.7% |
| Lineage coverage (tier 1) | 100% | 85% | ↑ from 60% |
| Owner coverage (tier 1) | 100% | 100% | flat |
| Median resolution (days) | ≤ 7 | 5 | ↓ from 11 |
How to start without a tool
- SQL views to compute accuracy and completeness on the top 3 data sets
- Power BI / Looker dashboard refreshed monthly
- One-page narrative summary for leadership – numbers do not speak for themselves
When to invest in a data quality platform
Above 20 in-scope data sets, or when regulators (FCA, NHS DSPT, ICO investigations) need formal evidence, a platform pays for itself. Below that, native tools and discipline usually win.
Targets are sector-specific
Banking targets are tighter than charity targets. Set bands that reflect your risk appetite and your customers' expectations. Publishing the targets matters more than the exact numbers.
How DQ KPIs connect to AI
AI built on data without these KPIs is a guess at scale. The data foundationspiece sets out the connection. Use the KPIs to evidence AI readiness in your impact assessments.