Running DMAIC Well in a Volatile Year: A Side-by-Side Look
DMAIC — Define, Measure, Analyze, Improve, Control — is the backbone of most Lean Six Sigma work. The five letters are easy to recite. Running them well is harder, and the gap between a project that sticks and one that fizzles usually shows up phase by phase. With 2022 bringing inflation, scarce labour and materials, and a return-to-office shuffle, teams are under pressure to fix things fast. That pressure is exactly when the discipline of DMAIC pays off — or gets skipped.
Below is what good and bad look like at each step. None of this is exotic; it is the difference between treating DMAIC as a habit and treating it as a slogan.
Define and Measure: aim before you fire
A good Define phase produces a tight problem statement, a measurable goal, a scope you can defend, and a sponsor who has actually committed resources. The team can say in one sentence what is wrong, for whom, and by how much. A bad Define phase opens with a solution already chosen — "we need new software" — and a problem statement so broad it could mean anything.
Measure is where good projects get honest. Strong teams confirm the data is trustworthy before they analyze it, often with a quick measurement-system check, and they baseline the current performance so improvement can be proven later. Weak teams grab whatever numbers are handy, never ask whether the data means what they think it means, and skip the baseline — which makes any later claim of success unfalsifiable.
Good: one-sentence problem, baselined metric, committed sponsor, validated data
Bad: pre-chosen solution, vague scope, borrowed numbers, no baseline
Analyze and Improve: find the cause, not a culprit
The Analyze phase is where rigour separates real improvement from guesswork. A good team uses the data to find the root cause and tests its theory before acting — five whys, a Pareto view, a process map walked in person, or a simple comparison of when the defect appears and when it does not. A bad team jumps to the cause that confirms what they already believed, often a person rather than a process.
Improve should follow logically from Analyze. Done well, the team trials the change on a small scale, measures whether the metric actually moves, and resists the urge to bolt on five other ideas at once. Done badly, the team implements everything immediately, changes several variables together so nobody can tell what worked, and declares victory the moment things feel better.
Verify the cause. Use evidence, not the loudest opinion in the room, and test the cause-and-effect link before you commit.
Pilot, then scale. A small controlled trial tells you whether the fix works before it is rolled out everywhere.
Change one lever at a time. If you alter five things at once, you will never know which one mattered.
Control: make the gain stick
Control is the phase most often shortchanged, and it is where the savings either become permanent or quietly evaporate. A good Control phase updates the standard work, sets up a simple monitoring signal so a drift gets caught early, hands ownership to the people who run the process every day, and closes out with a short record of what changed and why. A bad Control phase ends at the celebration email; six months later the process has crept back to where it started and nobody noticed.
In a year of supply shocks and turnover, that last phase matters more than usual: gains you cannot hold are gains you will have to win twice. The teams that come out ahead in 2022 are not the ones with the fanciest tools — they are the ones who finish the cycle.
If your team needs help turning an improvement effort into a result that lasts, XNM's strategic advisory can help you frame the problem and hold the gain.