Every year, poor data quality costs organizations an average USD 12.9 million. Apart from the immediate impact on revenue, over the long term, poor quality data increases the complexity of data ecosystems and leads to poor decision-making.
The emphasis on data quality (DQ) in enterprise systems has increased as organizations increasingly use data analytics to help drive business decisions. Gartner predicts that by 2022, 70% of organizations will rigorously track data quality levels via metrics, improving it by 60% to significantly reduce operational risks and costs.
Data quality is directly linked to the quality of decision-making. Good quality data provides better leads, a better understanding of customers, and better customer relationships. Data quality is a competitive advantage that D&A leaders need to improve upon continuously.
No. 1: Establish how improved data quality impacts business decisions
Identify a clear linkage between business processes, key performance indicators (KPIs), and data assets. Make a list of the existing data quality issues the organization is facing and how they are impacting revenue and other business KPIs. After establishing a clear connection between data as an asset and the improvement requirements, data and analytics leaders can begin building a targeted data quality improvement program that clearly defines the scope, the list of stakeholders, and a high-level investment plan.
No. 2: Define what is a “good enough” standard of data
To improve data quality, first, it is important to understand what is the “best fit” for the organization. This responsibility of describing what can be defined as “good” lies with the business. Data and analytics (D&A) leaders need to have periodic discussions with business stakeholders to capture their expectations. Different lines of business using the same data, for example, customer master data, may have different standards and therefore different expectations for the data quality improvement program.
No. 3: Establish a DQ standard across the organization
D&A leaders need to establish data quality standards that can be followed across all business units in the organization. It is likely that different stakeholders in an enterprise will have different levels of business sensitivity, culture, and maturity, so the manner and speed with which requirements of DQ enablements are met may differ.
This will enable stakeholders across the enterprise to understand and execute their business operations in accordance with the defined and agreed-to DQ standard. An enterprise-wide DQ standard will help educate all involved parties and make the adoption seamless.
No. 4: Use data profiling early and often
Data quality profiling is the process of examining data from an existing source and summarizing information about the data. It helps identify corrective actions to be taken and provides valuable insights that can be presented to the business to drive ideation on improvement plans. Data profiling can be helpful in identifying which data quality issues must be fixed at the source, and which can be fixed later.
It is, however, not a one-time activity. Data profiling should be done as frequently as possible, depending on the availability of resources, data errors, etc. For example, profiling could reveal that some critical customer contact information is missing. This missing information may have directly contributed to a high volume of customer complaints and would make good customer service difficult. DQ improvement in this context now becomes a high-priority activity.
No. 5: Design and implement DQ dashboards for monitoring critical data assets, such as master data
A DQ dashboard provides a comprehensive snapshot of data quality to all stakeholders, including data from the past to identify trends and patterns that can help design future process improvements. It can be used to compare the performance over time of data that is critical for key business processes. This enables the organization to make the right business decisions to achieve the desired business objectives based on trusted quality data.
DQ dashboards also reflect the impact of improvement activities, such as incorporating new data practices into operational business processes. They can be customized to meet the specific needs of a business and it shows how much trust you can put in your data.
No. 6: Move from a truth-based semantic model to a trust-based semantic model
The source of data is not always internal, where data quality can be controlled and maintained right from the beginning. In some cases, data assets are acquired from external sources where the DQ rules, authorship, and levels of governance are often unknown. Hence, a “trust model” works better than a “truth model.”
This means that, rather than thinking about key enterprise data as being absolute, organizations must also consider its origin, jurisdiction and governance — and therefore the degree to which it can be used in decision making. D&A leaders can implement mitigation measures when trust levels are not maintained.
No. 7: Include DQ as an agenda item at D&A governance board meetings
D&A leaders need to link DQ initiatives to business outcomes, which will help track the investments in DQ improvement against the business objectives. To get the board’s attention, it is important that the impact of DQ improvement is communicated to the board in a language they understand best — business and revenue impact. The board needs to have clear visibility of the DQ improvement progress and challenges, and they need to get this information on a regular basis.
No. 8: Establish DQ responsibilities and operating procedures as part of the data steward role
A data steward is responsible for ensuring the quality and fitness for purpose of the organization’s data assets, including the metadata for those data assets. In more mature organizations, a data steward’s role is also to champion good data management practices, and monitor, control or escalate DQ issues as and when they occur.
D&A leaders need to include this role in their D&A strategy so that DQ is measured and maintained regularly in a systematic manner. Create a governance scope and stakeholder map that will allow a clear understanding of how DQ issues are managed.
No. 9: Establish a special interest group for DQ across BUs and IT, led by the chief data officer team or equivalent body
A dedicated group that has representation from BUs, IT, and the office of the CDO that collaborates for DQ improvement can be a great investment of time and resources. Such collaboration enables better organizational management of risk. It also creates more opportunities for reducing operational cost and encourages growth through shared and consistent best practices.
No. 10: Establish a DQ review as release management “stage gate”
Review and update progress to make timely corrections and checks. As the organization’s maturity to handle DQ initiatives improves, identify and circulate the best practices that have been impactful.
No. 11: Communicate the benefits of better DQ regularly to business departments
D&A leaders need to measure the impact of the improvement program and communicate the results periodically. For example, a 10% improvement in customer DQ can be linked to a 5% improvement in customer responsiveness, since customers can be serviced better and faster by customer care executives due to the availability of good-quality, trusted data.
It is not only important to have the board’s attention in DQ improvement, but also for it to be a sustainable practice. It is important that the benefits are communicated to the board periodically.
No. 12: Leverage external/industry peer groups, such as user groups from vendors, service providers, and other established forums
D&A leaders can connect the enterprise with DQ peer groups and encourage organizational maturity in this area. This will enable them to exchange alternative perspectives on best practices and insights into the approaches others are taking to address similar challenges.
The original article by Melody Chien, senior director analyst at Gartner, is here.
The views and opinions expressed in this article are those of the author and do not necessarily reflect those of CDOTrends. Image credit: iStockphoto/DKosig