Putting Data Quality at the Forefront of Your Data Governance Policy

Data is of little use to an organization if it doesn’t have an appropriate level of quality. In order to achieve this aim, it must be at the forefront of your data governance policy. Otherwise, you’ll find that what you’ve collected can’t be used for its intended purpose.

How does data quality and data governance fit together?

Robust data quality practices ensure data governance policies can be firmly adhered to and must be considered together. Allowing the two elements to work symbiotically making sure quality goals are at the heart of your data governance strategy will prevent limiting issues from arising in the future. These could include:

• Inconsistent data
• Data that contains errors
• Data in the wrong format, which can't be converted into a workable configuration

What data quality policies should a senior manager implement into their data governance framework?

Senior managers should set out clear guidelines for collecting data to protect its quality. Establishing these from the beginning of the process will lead to an efficient framework that gives employees confidence in the data they’re presented with.

1. Prohibit manual data entry

Inputting data manually opens up the system to the potential for human error. An automated solution should always be used for data collection wherever possible and your company’s policies should ensure this isn’t overridden at any point, no matter how well meaning the employee.

2. Opt for open standards over proprietary databases

Choosing open standards over proprietary databases overcomes the issues of formatting or translating hurdles. Instead, you’ll have the flexibility to choose whichever tools you wish to scrutinize your data. These can vary from business analytics and reporting software to learning machines, allowing your data to work harder for you.

3. Control data access

Access to data should be limited to those who really need it, as an excessive number of employees with the ability to modify it can lead to issues. These include security complications and an inconsistent approach, which ultimately reduces the quality of the data. Defining a list of those who require access in advance will save time in the long run.

4. Document your data governance framework

It’s one thing working out a data governance framework, but it’s another to ensure it’s presented in a written form that everyone can adhere to. All processes should be documented and procedures outlined so an external representative could understand the workings without prior knowledge. Any relevant code should be included in the documentation with notes attached to explain its uses.

5. Turn your workforce into citizen data scientists

You’re likely to have a limited number of data management experts within your organization, but there’s no reason why you can't empower your entire body of staff to be confident with data. Educating your workforce in the basics of data management will turn them into citizen data scientists and provide a foundation for high-quality data throughout your business.

6. Perform data integration in the cloud

Opting for cloud-based solutions to data storage and analysis challenges is an effective way to integrate your data. Among the benefits of storing your data in the cloud is the potential to scale it easily, while many such solutions come with powerful analysis tools. You’ll find the functionality outstrips any other option available at present.

7. Choose an appropriate integration tool

A good integration tool should be able to both read and write your data to carry out transformation and cleansing processes. This will allow it to improve your data quality through sorting, filtering, aggregating, deduplication and data matching. It must be able to do this across all of your distinct data sources.

8. Program your algorithms to recognize international data sets

Analytics tools don’t innately know the difference between languages or units meaning systems must be configured to deal with them. Failing to do so can lead to incorrect conclusions being drawn from the data, as can be seen in this example from NASA, which failed to recognize the difference between imperial and metric units. Even special characters and accents can have an impact on results if machines aren’t taught to identify them.