Gartner defines data quality as the usability and applicability of data used for an organisation’s priority use cases [1] – simply: is the data fit for its intended purpose?
Data quality is just one aspect of an organisation’s data management strategy. However, its importance is often overlooked and regarded as a mere technical detail.
Data is essential to how an organisation operates and is the foundation of digital transformation strategies such as Machine Learning and Artificial Intelligence. As organisations progress towards becoming data-driven, data quality becomes a critical and necessary requirement.
The lack of data quality in an organisation results in data and analysis errors, bad operational or strategic decisions, manual data handling, long data access times, and long and costly project implementation, thus impacting customer experience, reputation, regulatory risks and opportunity for new markets [2].
With suitable levels of data quality, organisations can improve operational efficiency, make informed decisions, and improve market concentration. Data quality is typically indicated by data quality dimensions.
Gartner states the nine commonly used data quality dimensions as Accessibility, Accuracy, Completeness, Consistency, Precision, Relevancy, Timeliness, Uniqueness and Validity [1].
Other data quality dimensions have also been proposed for use, including Compliance, Confidentiality, Credibility, Currentness, Efficiency, Integrity, Portability, Traceability, Understandability, and Recoverability [3].
There are various software tools available for data quality monitoring, or organisations can choose to develop custom solutions.
It is important that each organisation understands what the acceptable level of data quality is that is required for its operations, services and products, how it can be measured and what actions can be taken to improve it.
The meaning of data quality varies across different industries and individual organisations, as well as the use case or objective of concern.
Example 1
A manufacturing process company is likely to focus on accuracy if the amount of product is the case. Incorrect measurements can result in insufficient product for sales directly impacting revenue and customer trust.
Here, the accuracy of the product volume can be measured by applying profiling analysis or completing verification analysis of random samples. Actions for improvement include ensuring that measurement systems are functioning and maintained.
Example 2
An e-commerce company is likely to focus on timeliness if product inventory levels are the case. Incorrect inventory levels can result in overselling products, impacting financial operational activities, and customer satisfaction.
Timeliness can be measured via the time difference between inventory levels updating and purchase activity. Actions for improvement include optimising the systems’ update rules.
Implementation of data quality strategies can quickly become overwhelming and costly. This has potentially served as a hindrance to data quality strategies being adopted.
However, to realise the full benefit of becoming data-driven and successfully implementing digital strategies, organisations must embark on a data quality journey.
It is recommended that organisations apply a practical approach to data quality – identify the critical business use cases for data quality, implement data-appropriate quality metrics, and then take necessary actions for improvements.
References
1. Data Quality: Why It Matters and How to Achieve It (https://www.gartner.com/en/dataanalytics/topics/data-quality)
2. Quality Management in Data Governance (https://www.deloitte.com/ce/en/services/consulting/perspectives/bg-qualitymanagement-in-data-governance.html)
3. A Framework for Current and New Data Quality Dimensions: An Overview (2024, Miller, R.; Whelan, H.; Chrubasik, M.; Whittaker, D.; Duncan, P.; Gregorio, J.)