Identifying the root causes of poor data condition
Good data is generally acknowledged as ‘valuable’, and where it’s not good, people agree that it’s worth fixing. So why measure data condition? Why not just fix it?
The purpose of conducting data condition assessments is not just to measure the gap between the data you have and the data you need, but to identify the root causes of poor data condition. A doctor wouldn’t medicate or operate without a proper diagnosis, nor should we blindly fix poor data when we don’t know the underlying issues. For that reason, we conduct qualitative and quantitative analysis to identify the root causes. The result: a remediation plan for lasting impact, and not just tactical fixes aimed at the symptoms.
So what are the key steps in Data Condition assessments? In our experience there are five:
1. Use case and capability definition
Use cases are high-level business activities, like maintenance planning or external reporting. Start by clearly outlining what an activity entails to define the scope of the assessment. Then elicit the capabilities required to fulfil each use case, both current and desired.
2. Critical data identification
Once use cases are defined, deduce the critical data required to enable each capability; this data may or may not exist within the organisation today. Don’t just assess against what currently exists, but against a blueprint for the data asset required to fulfil the use case.
3. Requirement elicitation
Critical data needs to be fit for purpose along several dimensions. For each critical data element, define whether and how it should be complete, valid, unique, precise, current or timely. Where the data exists, identify the relevant attributes in the systems of record and then define how validity, timeliness, precision etc. should be measured in relation to the specific use case. These items form your data requirements.
4. Technical assessment
This is where we measure the gap between the data you have and the data you need. Too often, data issues are anecdotal, people know the data is bad, but don’t know the scale of the problem. Having defined the requirements, we can measure and evidence the percentage of records that meet the specifications.
5. Root cause analysis
The results from the technical assessment provide an excellent basis to engage the business about their data. Identifying root causes of the issues we see in the data is done through conversations with people who understand the data and historic challenges. It is also good to do a technical root cause analysis, taking the ‘failing’ records and identifying trends that might indicate potential root causes.
In our experience, step one is the hardest, specifically: identifying the capabilities that don’t exist today, but could exist with better data. It’s all too easy to get stuck in the present and fail to devote time to creating a blueprint for a data asset that will level up your capabilities. It’s the hardest step, but it’s the most important. If you know what you’re aiming for, you can measure against it and work towards it.
What do you think? What challenges have you faced around improving the quality of your data?
Uncover the condition of your most critical data assets
Getting to the root of what’s affecting the quality of your data can be challenging, but we can help. Get in touch to discuss the ins and outs of doing a data condition assessment for your organisation.