Six Sigma: Improve Phase

Having successfully identified and verified the root causes of process defects in the Analyze phase, you now reach the pivotal Improve phase. This is where theoretical understanding transforms into tangible, measurable change. The Improve phase is the creative and rigorous engine of the DMAIC cycle, focused on generating, selecting, and implementing solutions that directly address root causes to achieve significant reductions in process variation and defects. Mastering this phase requires a blend of innovative thinking and disciplined methodology to ensure improvements are effective, sustainable, and aligned with project goals.

Generating Potential Solutions

Before selecting a solution, you must create a robust set of options. The goal here is ideation, not immediate judgment. Techniques like brainstorming, reverse brainstorming (considering how to cause the problem), and benchmarking against industry best practices are common starting points. However, the core intellectual tool for this stage is Design of Experiments (DOE), a systematic, statistical method for understanding the relationship between multiple input variables (factors) and the key output variable. Unlike changing one factor at a time, DOE allows you to efficiently explore how factors interact with each other. For instance, if your analysis revealed that both temperature and pressure affect product strength, a well-designed experiment can determine the optimal combination, not just the optimal setting for each in isolation.

A basic full-factorial experiment tests all possible combinations of factors and their levels. The analysis of variance (ANOVA) from a DOE provides mathematical evidence of which factors have significant effects, often summarized in a Pareto chart of effects. This moves solution generation from guesswork to a data-driven search for the process's optimal operating window. The mathematical model from a simple two-factor experiment might look like this, where $Y$ is the output, $\beta$ are coefficients, and $x$ are the factor settings: $$Y={\beta }_0+{\beta }_1{x}_1+{\beta }_2{x}_2+{\beta }_{12}{x}_1{x}_2+\epsilon$$ This equation helps you predict performance for any combination of $x_1$ and $x_2$.

Evaluating and Selecting the Best Solution

With a list of potential solutions generated through DOE and other creative methods, you must now choose the one to implement. This requires a balanced assessment of impact, feasibility, cost, and risk. A solution selection matrix (also called a Pugh matrix or decision matrix) is the primary tool for this objective evaluation. You list your candidate solutions against a set of weighted criteria, such as estimated defect reduction, implementation cost, time to implement, and ease of maintenance. Each solution is scored against the criteria, and the weighted scores are summed to provide a comparative ranking.

Concurrently, a Failure Mode and Effects Analysis (FMEA) is conducted on the top candidate solutions. This proactive risk assessment tool systematically identifies everything that could go wrong with a proposed solution (failure modes), the severity and likelihood of those failures, and the ability to detect them before they impact the customer. You calculate a Risk Priority Number (RPN) as: $RPN=Severity(S)\times Occurrence(O)\times Detection(D)$ The FMEA process forces you to design countermeasures into the solution itself, making it more robust. The output from the selection matrix and the FMEA are reviewed together, often leading to the modification or strengthening of the top-ranked solution before it moves forward.

Piloting the Solution

A full-scale, irreversible rollout of an untested solution is a high-risk proposition. Therefore, the piloting of the selected solution is a non-negotiable step. A pilot is a small-scale, controlled test of the improvement under real-world conditions but within a limited scope—perhaps on a single production line, one shift, or with one client team. The objectives of a pilot are to: verify the predicted performance benefits in practice, identify unforeseen implementation challenges, assess training needs, and validate the updated process controls and measurements.

During the pilot, you collect the same key performance data (like process capability indices, defect rates, or cycle time) that you used in the Measure and Analyze phases. This creates an apples-to-apples comparison to the baseline. A successful pilot provides not only proof of concept but also a refined implementation plan, as you learn what works and what doesn't in a low-consequence environment. It is your final quality check before committing organizational resources to a full launch.

Planning and Executing Full Implementation

A successful pilot sets the stage for the final, full-scale implementation. This requires detailed implementation planning that mirrors strong project management principles. Your plan must address the who, what, when, where, and how of the rollout. Key elements include a communication strategy to secure buy-in from all stakeholders, a comprehensive training plan for those who will operate the new process, a timeline with clear milestones, and a definition of the new standard operating procedures (SOPs).

Crucially, the implementation plan must also detail the new control plan. This document specifies how the improved process will be monitored and maintained long-term. It answers questions like: What are the key metrics to track? How often will they be measured? Who is responsible for reviewing them? What are the response plans if metrics drift from their improved levels? Integrating the solution into the organization's formal management systems ensures the gains are locked in and sustained, creating the bridge from the Improve phase to the final Control phase of DMAIC.

Common Pitfalls

1. Jumping to the First Solution: The most common mistake is bypassing structured ideation and evaluation. Selecting a familiar or seemingly obvious solution without using a selection matrix or FMEA often leads to suboptimal results or unintended consequences because interactions and risks were not fully considered.
2. Skipping the Pilot: Teams under pressure may try to move directly from a paper solution to full implementation. This almost always leads to catastrophic rollout failures, employee resistance, and wasted resources. The pilot is your essential "dress rehearsal" that uncovers practical roadblocks.
3. Neglecting the People Side: Focusing solely on the technical solution while ignoring change management is a critical error. Failure to communicate the why, provide adequate training, and address concerns will lead to poor adoption, even if the solution is technically perfect. People execute processes, and their buy-in is essential.
4. Incomplete Implementation Planning: Launching a new process without a documented control plan, updated SOPs, and assigned accountability guarantees that the improvements will degrade over time. The work of the Improve phase is wasted if the new state is not systematically institutionalized.

Summary

• The Improve phase is the action-oriented core of DMAIC, transforming root cause analysis into validated, implemented solutions that reduce variation and defects.
• Design of Experiments (DOE) provides a statistical framework for efficiently generating and optimizing solutions by understanding how multiple input factors interact to affect the output.
• Solutions must be rigorously evaluated using a solution selection matrix for balanced scoring and a Failure Mode and Effects Analysis (FMEA) for proactive risk assessment and mitigation.
• Always pilot the chosen solution on a small scale to confirm benefits, refine the approach, and identify unforeseen issues before a full, resource-intensive rollout.
• Successful, sustainable implementation requires a robust plan covering communication, training, updated procedures, and—most importantly—a detailed control plan to monitor and maintain the new process performance.