The Powerful Combination of Improving your Data and Processes Together

Early in my data career I attended a keynote at a local conference and the speaker's key message to the audience was that behind the data there are always people. That message stuck with me as I got more experience with data and found it 10 times easier (at least!) to navigate the data when I understood how it came to exist. To go with that approach, experience also taught me which questions to ask depending on what those data-generating-processes looked like. This is one perspective on how analyzing your processes can support your data projects.

In Lean Process Improvement, we rely on people most of the time to identify the inefficient processes and prioritize which ones should be eliminated. Time studies are one source of data we use to be more accurate in those process improvement efforts. We also sometimes use operational data to help us better quantify the scale of a specific problem. See a relevant xkcd webcomic for how those elements of information can help you decide what is worth the effort:

It may be immediately clear why data projects can benefit from process understanding and additionally process improvement projects benefit from supplemental data. However, the really powerful combination are Data+Process projects. A data project promises to help an organization have easier access to useful data, which often is limited by the data-generating processes upstream of how the data is intended to be used. Similarly, process improvement projects may be limited by the quality of the data available, or at least it is challenging to identify the most critical improvement opportunities without a quantification of the scale of the problem.

A Data + Process problem can take many forms. Here are a few examples:

• Manual data entry because no one created a set of choices in the system. Or they created a set of choices with overlap and gaps so the data is difficult to interpret.

• Some inventory items are purchased outside of the normal process because they are not stocked on hand, resulting in limited information about what was purchased, preventing it from being stocked in the future.

• Processes are delayed or include rework because of missing data.

• Processes that should be automated are manual to fill in gaps in data flows.

Solving Data + Process problems is often more complicated than one or the other. In mathematics you learn about "degrees of freedom" and that when you have more options, there is less clarity in what the right path may be. However, while it may be hard, solving Data + Process problems from both angles result in significantly improved outcomes. Any problem that can be solved with only one domain still could be, but there are a host of problems that can only be solved with a combination of the two approaches. And by considering both, you can identify the highest ROI solution to your problem.