“Data Quality; unlocking the potential of your data to reveal quick wins or areas for achieving rapid payback”
Damien Smith of Adelo was on next in Stream Two (unlocking the value from your data). His presentation was focused on quality issues, rather than quick wins or rapid payback, but was still relevant and well received.
Much of Damien’s session centred on the themes of poor quality or data that was not understood and the ongoing tales of I.T. developing workarounds rather than the business tackling the underlying problems. Damien’s strategy was to propose a “Data quality management service” for the organisation with a remit to focus on Data Quality and Data Profiling.
Data Profiling
- What’s the standard?
- Does data conform to expected levels?
- Are there implied standards if no standards exist?
- Are the standards good enough?
Back in September I spoke at the Obis Omni forum on developing a Developing_BI_CPM_Strategy. I recounted a conversation I had with Chris Hooker of Infor back in the early days of our Data Warehouse project. Chris told me that in his considerable experience as a consultant the biggest issue he encountered was data quality. He found business were surprised to be told they had data issues, often being unwilling to accept the hard truth.
From my perspective data quality is an ongoing priority, especially as I work for a global company with some distinct ideas on what constitutes a “Standard”. There no doubt in my mind that Damien has a point. Create a standard and live by it.
Data Quality
So you have a standard in place but how should be alerted to issues and what should be done about them?
Damien spoke a lot about “IT” attempting to solve the problems of the world rather than seeking business sponsorship to resolve them. It’s a familar and worrying theme. He suggested storing issues centrally to give them life beyond a project and to implement ongoing reporting throughout the business.
It’s been a few weeks since the conference, but under his presentation in my note book I have four bullets, titled “Do Tomorrow”. They are:
- Start recording issues and escalate
- Start reporting data quality
- Understand responsibility for cleansing data – especially outside of the project environment
- Gain sponsorship for a Master data management process
Thanks Damien, good advice and wise words.