Data Quality is Critical – Key Actions for Project Managers

Poor data quality can impact virtually every phase of a project. I've had instances where the cutover was a risk because of data quality issues. Data quality is usually overestimated; please keep this in mind when planning and managing a project. Let's look at what you need to do to manage this risk.

Why is it so difficult to determine data quality?

In many organisations issues with data quality are hidden. Key stakeholders you meet with during project initiation may not be aware of the problems. I've heard statements like, 'I'm sure the data quality is fine because we aren't seeing BAU issues. We're using this data to run the business.'

The senior managers may not be aware of the additional processes that have been put in place to cope with the data inconsistencies. This, unfortunately, is common.

What are the reasons for poor data quality?

There are several reasons for poor data quality including lack of ownership, issues with manual data input, lack of validation and poor review and update procedures.

Data quality 

Data quality is critical. This rule applies to every project regardless of size or methodology. Quality assurance is required for all activities, but data quality assurance is an area that requires additional diligence. 

A few thoughts on data quality:

• If data isn’t updated, cleansed and quality assured it can cause problems with testing, dress rehearsals, cutover and BAU.
• In many scenarios, test data must match production data. If you don’t have quality data, there is a high probability that the problem will impact all test phases.
• Knowing that your data is as close as possible to 100 per cent before go-live is critical for post-go-live support.
• An example from a project. After cutover, issues arose with one of the upstream applications. This application was part of BAU before go-live and was not part of the implementation.
• There were multiple upstream systems, and troubleshooting the issue was difficult.
• The data for the new application underwent rigorous data assurance and was confirmed to be accurate before the cutover.
• The level of data assurance allowed the team to pinpoint the issue in the upstream system quickly because they had detailed information on the data in the new application. Without this solid data baseline, it would have taken far longer to resolve this issue, and BAU would have been impacted.
• Always consider the multiple uses the data will have.
• You may be working with the manager of one of the departments in finance, and the data he is seeing in the tests is correct. However, the finance director may have other reports that consolidate data or use a different data set. Include scenarios of this type in the test and dress rehearsal activities.
• Problems with data can cause reporting issues. (Inevitably the issues will be seen in the reports the senior stakeholders use.) 
• This situation can cause stakeholders to question whether the application is functioning correctly, despite it actually being a data issue.
• Don’t make any assumptions on data quality.
• When asked about current data quality, a manager on one project said, ‘The data quality must be okay; they are using it for BAU.’
• What the manager didn’t know was that the team had a process in place to manage the issues with the data.
• The data quality was very poor, and updating the customer information took several weeks.

In summary, ensure you do everything possible to have complete data accuracy before cutover. Data is the lifeblood of an organisation, and high-quality data is critical for cutover and BAU.