Honest data that misleads

"In God we trust. All others must bring data." - W. Edwards Deming

Deming's quote isn't just a quip; it holds weight. Firstly, he's my idol in the realm of management systems, and secondly, I'm still astonished by how what he described 60 years ago is still conveniently forgotten and underutilized. In this piece, I aim to address a critical issue regarding data handling because I still see data being compared year to year, month to month, and quarter to quarter. I even witnessed comments drawn from such data, which were conclusions about positive or negative trends. From a data analysis standpoint, let alone assessing a company's performance, this approach needs to be revised. Let's delve into why.

Imagine data presented as follows, as is often the case in media:

Headline 1: UK Trade Balance Deficit Decreases by 8%

Headline 2: UK Trade Balance Deficit Increases by 35%

Headline 3: UK Trade Balance Deficit Increases by 8%

Headline 4: UK Trade Balance Deficit Decreases by 4%

The first question that arises is: What does each headline tell us? Is it good or bad? Can we say anything substantial based solely on this data? This method of presenting data only informs us about changes, but what specific changes, and over what period?

Let's try to enhance the data by presenting it like this:

Headline 1: UK Trade Balance Deficit Decreases by 8%, totaling ￡7.1 billion

Headline 2: UK Trade Balance Deficit Increases by 35%, totaling ￡9.6 billion

Headline 3: UK Trade Balance Deficit Increases by 8%, totaling ￡10.4 billion

Headline 4: UK Trade Balance Deficit Decreases by 4%, totaling ￡9.9 billion

This expression provides more specifics, and one can even draw some conclusions. But what can we say about the overall system's condition? Is it still good or bad? Over what period are we talking? Because we might also see a headline like this: Relative to October 2020, the UK trade balance deficit in October 2021 decreased by 0.5%.

This seems like good news, but what if we step back and look at the big picture?

The section highlighted in yellow represents the "news' headlines" that were reported, while October 2020 and 2021 are highlighted in red.

Now, we can definitively conclude where the process is heading and what's happening with it. All values are used in context and complement each other. With this data, we can start working. And even if quarterly comparison data suggests that business is going well, deeper data might reveal a negative trend emerging or already present within weeks.

Moreover, if we had predefined process boundaries, precisely calculated ones, not arbitrarily assigned ones (because methods exist for calculating process behaviour boundaries), we would have detected issues with the current process even earlier and taken action to address the causes of process variability.

Interestingly, given the current organization, the process flows as orderly as possible, but changing it to make it more manageable or to steer it in the desired direction already requires managerial decisions rather than operational ones. But that's a topic for another conversation.

The conclusion is simple: managing a company/process/operation is based on factual data about the process's progress. Everything else is driving a car forward while looking in the rearview mirror, as one author whose name I can no longer recall said.

Next time you see a comparison of one year's quarter to another (for sales, marketing campaigns, goods manufacturing, quality issues, etc.), remember that these data points are critically insufficient for making managerial decisions because they do not reflect the system's behaviour. If you are still in doubt, ask yourself why Bloomberg shows the deepest and most historically complete data level it knows to help make decisions.

Data for the chart was taken on www.ons.gov.uk

#data, #datamanagement, #management, #process, #progress, #quality, #decisionmaking, #decision