In December, I wrote about the Measure Phase of DMAIC, emphasizing the importance of collecting facts and aligning around them so we can collaboratively work on a problem. This month, I will focus on the Analyze phase, where those facts are organized to generate meaningful insights. During the Analyze phase, we develop a deeper understanding of the causes behind the effects we have measured. While many tools are introduced in this phase, rather than focusing on the tools themselves, I will explain how to think about the Analyze phase of DMAIC and the mindsets that are essential to finding the best solutions.
Let’s talk a bit about math. I recently started reading a book that has helped me think more clearly about the role math plays in the Analyze phase. Unfortunately, most of us were taught math as a collection of symbols and procedures for manipulating numbers. That approach often leaves us with the impression that math is boring at best and an excruciating mental exercise at worst. So what’s the point if I don’t plan to build bridges, right?
In reality, math is essential to how we establish cause-and-effect relationships. It is a method we use to abstract and conceptualize the real world in our minds. By doing so, we gain an objective understanding of causal relationships. Without math, problem solving becomes my opinion and anecdotes versus yours… or worse, my power of personality against yours. In the Analyze phase, what we are truly seeking is an objective understanding of cause and effect.
Consider an early human trying to throw a rock, perhaps while running from a charging animal. Do I pick up the flat rock? The large rock? No, the round rock. The one with a circumference that fits comfortably in my hand and can be thrown accurately at the animal’s head. The rock that will deliver enough force to disrupt the charge. Math helps explain this simple relationship: the cause (the size, shape, and weight of the stone) produces the effect (how accurately and forcefully it can be thrown).
There is another important concept embedded in the Analyze phase. As modern humans, we constantly make these kinds of cause-and-effect judgments. Is that box too heavy for me to lift? How hard do I need to press the brake to stop at the sign? We aren’t consciously doing calculations, but at its core, this is mathematical thinking. The challenge is that while we have an instinctual sense (our intuition) about the right answer, that intuition is not always correct.
When we quantify cause-and-effect relationships using math, we create a feedback mechanism. That feedback allows us to objectively evaluate outcomes as good or bad, right or wrong. In this way, analysis becomes a “check” on our intuition and over time, it improves it. The more we study and quantify cause and effect, the stronger our intuition becomes. Stronger intuition allows us to find and fix problems faster. This virtuous cycle is Continuous Improvement, quietly at work within the Analyze phase.
Hopefully, those of you who have worked with me can now see why I push so hard against jumping to solutions. Quick solutions are intuition at work. Intuition is powerful and valuable, but without the feedback mechanism of analysis, it is little more than guesswork. Worse, individual intuition rarely aligns across a team. The greater the differences in experience, the greater the differences in intuition. The result is disagreement about solutions, friction over next steps, and stalled progress. Teams need the aligning force of a shared, data-based understanding of cause and effect provided by math.
Now let’s connect the Measure and Analyze phases to complete the picture. In Measure, we collect facts. Doing so requires math to abstract reality into a form we can document and discuss. This quantification allows us to ask how variables interact to produce the effect of interest. In Analyze, we examine the impact of causes on effects (using numbers, graphs, and models) to validate that relationships truly exist. This analysis informs and sharpens our intuition, enabling us to creatively identify ways to control those relationships. That work leads us into the Improve and Control phases. The tools we teach (charts, graphs, and mathematical models) serve as documentation that aligns the team and educates our collective intuition, so we move forward together when it’s time to improve.
Let’s bring this to life with a simple example. Imagine a room full of experienced professionals. A salesperson complains that a customer is not receiving product consistently on the promised date. The operations manager says the issue is late orders arriving at the distribution center. The IT leader claims the conveyor systems are always broken. The maintenance leader insists the operators are not taking care of the equipment. These are all uninformed intuitions, each shaped by their own experiences.
Now consider the same discussion after analysis. The salesperson reports that on-time delivery for product ABC is 95%, despite a customer promise of 99%. The operations manager notes that orders are being released at 10 a.m. which is one hour later than the 9 a.m. service-level expectation. The IT leader responds, “I wasn’t aware of that; I’ll investigate,” and adds that the conveyor system frequently goes down, causing missed 3 p.m. shipping cutoffs. The maintenance leader reports that conveyor uptime is 90%, well below the 99% target, and that lubrication records show required daily procedures are being skipped three out of five days each week. The operations manager acknowledges the need to investigate this procedural slippage.
This story is not hypothetical. In my 25 years working in process improvement, I have witnessed both versions of this meeting. In both cases, people relied on intuition. But in the first, intuition was poorly informed. There was no math, no analysis, and no feedback mechanism. In the second, intuition was informed by data and analysis, and the team improved their shared understanding of the process. Over time, this leads to more robust and productive problem-solving discussions. That is the mindset and the power of the Analyze phase.