How to Improve a Process with One Tool from Every DMAIC Stage

DMAIC — Define, Measure, Analyse, Improve, Control — is the backbone of every Six Sigma project. But it is easy to learn the five words and still have no idea what you would actually do on a Monday morning. So here is a full walkthrough of one improvement project, using exactly one tool per stage, in a setting everyone understands: a packing department. The scenario and numbers below are an illustrative example of a typical small packing operation, so you can see how the pieces connect and how the maths behaves.

The Packing Department Scenario

A small online retailer’s packing department has three staff, one bench and a growing reputation for slow dispatches and mislabelled parcels. Orders wait in totes, walkers criss-cross the room for boxes, tape and labels, and nobody can say with confidence how long a typical order takes. The supervisor has been asked to “just make it faster”. Instead, she runs a DMAIC project.

Define: Project Charter

The project starts by writing a Project Charter: one page that pins down the problem, the goal, the scope, the team and the timeline. The charter states the problem as a measurable gap, not a mood: “Average pick-to-dispatch time is 46 minutes against a target of 25, and 3.6% of parcels are mislabelled.” The scope excludes the warehouse picking upstream and the courier downstream; the packing room is the project.

Alongside the charter, a quick SIPOC diagram sketches the process at arm’s length: picked orders come in, a label and a box go on, a sealed parcel goes out. Define ends when everyone agrees what “better” means and what is out of bounds.

Measure: Baseline Metrics and DPMO

Guesswork dies in Measure. The team builds a data collection plan and records, for two normal weeks, the start and finish time of every order and every defect type (wrong label, damaged box, missing insert). The raw numbers land in the baseline metrics template: average 46 minutes pick-to-dispatch, 3.6% mislabelling, 19 hours of overtime a week.

To make the defect rate comparable, they convert it with the DPMO calculator: one defect opportunity per parcel (correct label, correct box, correct insert), giving 36,000 defects per million opportunities, roughly a 3.2 sigma process. That single number is the project’s starting line, and it is measured, not felt.

Analyse: Pareto Chart and Fishbone Diagram

Two weeks of defect data produce a long, depressing list of causes — until it goes into a Pareto Chart. The 80/20 pattern is textbook: 79% of mislabels trace to two causes — the label printer sits across the room from the bench, and two similar product SKUs are distinguished only by a character buried in the label text.

For the waiting problem, the team runs a Fishbone Diagram and then drills the biggest bone with the 5 Whys. Why do orders wait? Because packing stops when an item is missing. Why is it missing? Because the pick list is not checked on arrival. Why not? Because there is no defined checking step. The trail ends at a missing standard, not at a lazy person. Analysis always aims at the process, never the people.

Improve: 5S and Poka-Yoke

Improvements go in cheapest-first. The team runs 5S on the room: the label printer moves next to the bench, boxes are sorted by size into one labelled flow rack, and every tool gets a shadow-marked home. A new arrival-check step, written as standard work, kills the missing-item waits.

For the SKU mix-ups they add a poka-yoke (mistake-proofing): the label software now prints the product name in 20-point text and a colour bar per SKU family, so a wrong label is visible before it sticks. None of these fixes is expensive; together they change the process.

Control: Control Plan and Control Charts

Improvements decay unless something holds them. Every fix is written into a control plan: who checks what, how often, and what to do when it drifts. Daily mislabel counts go onto a control chart, with limits calculated properly in the Control Limits calculator, so the team reacts to genuine shifts instead of normal noise. The supervisor keeps a before-and-after comparison on the wall: it is the project’s receipt.

Before and After: The Results

Four weeks of the new process, measured the same way as the baseline:

Notice what made the story credible: the team measured before they changed anything, attacked the two vital-few causes instead of everything, mistake-proofed rather than nagged, and kept measuring after go-live. That is all DMAIC is — a discipline for doing what most teams already try to do, but with evidence at every step.

Every tool in this walkthrough is free on SimplicityHub: the full DMAIC template set, the template library and the calculators page cover every stage from charter to control chart. To learn the method end to end, the free Green Belt course walks through DMAIC project by project.