Data quality
What it is, how to measure it and how to improve it: the full guide, a free calculator and what the AI Act requires.
What data quality is
Data quality measures how well your data fits the use you need it for. It is assessed through six dimensions: completeness, accuracy, consistency, timeliness, validity and uniqueness. Measuring it with simple indicators, rather than impressions, is the first step towards reports, decisions and AI systems that rely on trustworthy data. For high-risk AI, Article 10 of the AI Act also requires relevant and representative training, validation and testing data.
The six dimensions
- CompletenessNo required data is missing.
- AccuracyThe data reflects reality.
- ConsistencyThe same data matches across systems.
- TimelinessIt is available and up to date when needed.
- ValidityIt follows the defined format and rules.
- UniquenessThere are no duplicates.
Calculate your data quality score
Answer a few questions and get an indicative score per dimension, with no sign-up and in a few minutes.
Open the free calculatorFundamentals
Data Quality Dimensions: The 6 Core Dimensions Explained With Examples
What are data quality dimensions? The 6 core ones explained: completeness, accuracy, consistency, timeliness, validity and uniqueness, with practical examples and how to measure each.
Read →How to measure data quality: KPIs, thresholds and dashboards
Practical guide to measuring data quality with real KPIs, thresholds per dimension and how to build a quality dashboard the business actually uses.
Read →Data Quality Management: 2026 Guide (6 Dimensions + Checklist)
Data Quality Management explained with examples: the 6 dimensions, how to measure them, 2026 tools, and why the AI Act already requires it. Free practical guide.
Read →In practice
Data Profiling: What It Is and How to Do It Before Defining Quality Rules
What data profiling is, which techniques to use, and how to turn it into input for your data quality rules.
Read →Data Cleansing: 2026 Step-by-Step Process
The full data cleansing workflow: detection, standardization, deduplication, validation and enrichment. When to automate it and when it needs human review.
Read →Data Quality Tools in 2026: A Complete Comparison
Great Expectations, Soda, Monte Carlo, Collibra DQ and more: what each data quality tool does, when to choose open source vs SaaS, and how each fits your pipeline.
Read →Data quality and AI
Data quality for generative AI: what changes when you feed a chatbot your documents
When generative AI answers from your documents (RAG), their quality rules. What to check: currency, duplicates, permissions, sources and traceability.
Read →Training Data Quality: What Article 10 of the AI Act Actually Requires
What Article 10 of the AI Act requires about training, validation and test data in high-risk systems, with the updated timeline after the 2026 Digital Omnibus.
Read →Quick answers
Which data quality metrics should you use?
Which data quality metrics to use: one per dimension (completeness, accuracy, consistency, timeliness, validity and uniqueness), how to calculate them and how many are enough.
Read →What level of data quality is acceptable?
What level of data quality is acceptable: why there is no universal percentage, how to set thresholds based on the risk of use, and when to review them.
Read →Who is responsible for data quality in a company?
Who is responsible for data quality in a company: data owner, data steward and technical team roles, and how to split them in a small organisation.
Read →Where to start with data quality in a small business?
Where to start with data quality in a small business: pick one data set, measure it with a few metrics, fix the worst issues and assign an owner, without expensive tools.
Read →Frequently asked questions
What are the dimensions of data quality?
The six most used are completeness, accuracy, consistency, timeliness, validity and uniqueness. Each is measured with its own indicator and compared with a threshold set by the business.
How do you measure data quality?
For each important data element you define a rule per dimension, calculate the percentage of records that meet it and track the trend on a simple dashboard. The calculator on this page gives a first indicative score.
What is the difference between data profiling and data cleansing?
Data profiling analyses what your data looks like to uncover problems and the rules you need. Data cleansing fixes those problems: it standardises, deduplicates and validates.
What does the AI Act require about data quality?
Article 10 requires data governance practices for training, validation and testing data in high-risk AI systems: it must be relevant, sufficiently representative and, to the best extent possible, free of errors and complete.
Where to start in your company?
The free diagnostic shows where your data governance stands and what to prioritise.