SiliciumHex FieldKit

Problem Solving & Quality · Act

DMAIC Roadmap

Define, Measure, Analyze, Improve, Control: a data-driven project roadmap for chronic variation problems whose cause is not obvious.

  • Time8 h+
  • FormatTeam
  • StageAct

DMAIC Roadmap: what it is and why it works

DMAIC is the five-phase project roadmap of Six Sigma: Define, Measure, Analyze, Improve, Control. Define sets the problem, scope, customer requirement, goal and team in a project charter. Measure validates the measurement system, collects baseline data and quantifies current performance and capability. Analyze uses process mapping, stratification and statistical tools to find which input variables really drive the output. Improve generates, selects and pilots solutions that act on those proven drivers and confirms the gain with data. Control locks in the gain through a control plan, SPC, updated standards and a formal handover to the process owner.

DMAIC is built for chronic problems where the cause is not obvious and quick fixes have already failed: yield losses, variability, cycle time or defect rates that persist for months. Its discipline lies in the sequence: no solution before the drivers are proven, and no analysis before the data are trustworthy. Short gate reviews between phases catch shortcuts early. It is slower and heavier than PDCA or 8D, so it is overkill for problems with an evident cause. Within it, the other tools find their place: gauge R&R and capability in Measure, designed experiments and hypothesis tests in Analyze and Improve, the control plan and control charts in Control.

What you need

  • A chronic, measurable problem with a business case and a sponsor
  • A project leader with statistical skills and a cross-functional team
  • Access to process data, or the means to collect it
  • Time: typically several months of part-time work, with gate reviews
  • The customer requirement or specification that defines success

What you get

  • A project charter and a validated baseline of performance
  • Statistically confirmed drivers of the problem
  • Piloted and implemented improvements with a confirmed gain
  • A control plan, updated standards and a documented handover to the process owner

When to use it

When a chronic yield or variability loss has resisted quick fixes for months.

How to do it, step by step

  1. Define: write the project charter — problem, scope, customer requirement, goal, team, timeline.
  2. Measure: validate the measurement system, collect baseline data and quantify current performance and capability.
  3. Analyze: identify the input variables that drive the output, using process analysis, stratification and statistical tests.
  4. Improve: generate, select and pilot solutions that act on the proven drivers; confirm the gain with data.
  5. Control: lock in the gain with a control plan, SPC, standards and handover to the process owner.

Worked example: Batch yield variation on a resin reactor

Illustrative scenario — figures are realistic but not from a real company.

A specialty resins plant ran a 6,000-gallon batch reactor whose yield varied between 88 % and 94 %, averaging 91 %. Each point of yield was worth roughly $150,000 a year. Several quick fixes over a year, including catalyst supplier changes and operator retraining, had not moved the average. A green belt engineer led a DMAIC project with a sponsor from operations.

  1. Define: the charter targeted an average yield of 93 % with a standard deviation cut by half, within five months, scoped to one reactor.
  2. Measure: a gauge study of the yield calculation found that weighing of the product drums contributed large error; a scale was recalibrated and the weighing method standardized before collecting 40 batches of baseline data.
  3. Analyze: stratification showed no difference by catalyst lot or shift, but regression found that yield dropped when the monomer feed time exceeded 95 minutes, and feed time depended on the temperature of the cooling water.
  4. Improve: a designed experiment on feed rate and jacket temperature setpoint identified a combination that kept feed under 90 minutes in all seasons. It was piloted on 15 batches.

Result. Average yield rose to 93.2 % with a standard deviation cut from 1.6 to 0.7 points. Control: feed time was added to the control plan with an individuals chart and a reaction plan, and the process owner signed the handover. The team noted that without fixing the weighing first, the analysis would have chased noise.

Open it in the FieldKit Problem Solving app → Get the Problem Solving template pack — join the waitlist

Common pitfalls and how to avoid them

  • Moving to Analyze before the measurement system is validated.Hold a gate review at the end of Measure and require a gauge study on the key outputs and inputs.
  • Scoping the project too broadly, such as "improve plant yield".Narrow the scope to one product, line or unit with a measurable target that can be reached in a few months.
  • Jumping to a favorite solution in Define and using the phases to justify it.State the problem without a solution in the charter, and let the data in Analyze decide which drivers to act on.
  • Handing over the project without a working control plan.Close Control only when the process owner runs the monitoring and the reaction plan independently.

Frequently asked questions

What are the five phases of DMAIC?

Define the problem, scope, goal and team in a charter. Measure current performance, after validating the measurement system. Analyze the data to identify the input variables that drive the problem. Improve the process by selecting and piloting solutions that act on those drivers. Control the improved process through monitoring, standards and a control plan so that the gain is sustained.

When should you use DMAIC instead of 8D?

Use 8D when a specific incident or customer complaint needs a quick, traceable response, with containment and root cause. Use DMAIC for chronic, recurring performance problems such as yield, variability or cycle time, where the cause is unclear and several variables may interact. DMAIC takes longer and requires more data analysis, but it is better at solving problems that have resisted simple fixes.

What is the difference between DMAIC and DMADV?

DMAIC improves an existing process that is not performing well. DMADV, Define, Measure, Analyze, Design, Verify, is used to design a new process or product, or when an existing one needs a complete redesign rather than incremental improvement. Both are part of the Six Sigma toolkit; the choice depends on whether you are fixing something that exists or creating something new.

Origin

Six Sigma — Motorola (Bill Smith), 1986; the DMAIC roadmap was standardized during the 1990s (GE and others).

Related methods

More in “Act”

Plan, pilot and run the fix — with owners, dates and a structured roadmap.