A Toolkit for Process Improvement

Key ideas#

1. Setting priorities#

• Trend analysis: plot current measurements against past ones; a worsening trend marks a candidate.
• Spider chart: compare yourself with competitors on several measures at once; the largest gaps get priority.
• Performance matrix: plot processes by importance against current performance. High importance with low performance comes first. If that quadrant is empty, improve important processes that already perform well.
• Criteria testing: a matrix of processes against critical success factors. Score each process’s contribution to each factor (1–3), multiply by the factor’s weight (1–3), and add up per process. It moves attention from abstract success factors to concrete processes.

2. Understanding the problem#

• Critical incident: ask people which recent incident caused the most trouble, then sort answers by frequency.
• Check sheet: a simple form for counting events by category and period.
• Pareto chart: categories in descending order with a cumulative line. A few causes usually account for most occurrences.
• Flowchart of the current process.

3. Analyzing the problem#

• Cause-and-effect (fishbone) diagram and “5 whys”: see Root Cause Analysis.
• Scatter diagram: plot two variables (at least 30 points, ideally 100) to look for correlation.
• Histogram: the distribution of a measure in 5–20 classes. Two peaks suggest two mixed sources; a sharp cut-off at a limit suggests results being sorted or adjusted.
• Relations diagram: maps cause-and-effect links in complex problems. A factor with mostly outgoing arrows is a driver; one with mostly incoming arrows is an outcome.
• Matrix diagram: relates two or more sets of items, weighting each link as weak (1), medium (3) or strong (9).

4. Generating ideas and reaching consensus#

Brainstorming, written variants such as Crawford slip writing, nominal group technique and affinity diagrams.

5. Developing improvements#

• Simplification, following Harrington, in this order: remove bureaucracy (needless approvals, signatures, copies); remove duplication (the same data entered in several places); value-added analysis (classify each activity as adding value for the customer, adding value only for the business, or adding none, then remove the last kind); and cut cycle time (parallel work, better sequence, less waiting).
• Idealization: draw the ideal process from scratch, ignoring current constraints, and compare it with reality. The gap suggests solutions.
• Quality function deployment (QFD): a “house of quality” matrix that ranks improvement directions by what customers need.
• Work-cell analysis: analyze a person or unit with clear inputs and outputs. It uses the Kano model: stated requirements, basic requirements that customers expect without mentioning, and unexpected delighters. Meeting stated needs while missing basic ones does not satisfy anyone.
• Statistical process control: control charts with limits at three standard deviations separate chronic variation (many small causes) from sporadic variation (one clear cause). Investigate points outside the limits, long runs on one side of the center line, and trends.
• Business process reengineering: Hammer and Champy’s radical rethinking of processes for breakthrough gains in cost, quality, service and speed. Either systematic, starting from the current process and applying Eliminate, Simplify, Integrate, Automate in that order, or clean sheet, starting over from the customer needs the process must meet.
• Benchmarking: learning from others’ practices; the only tool with an external impulse.

6. Organizational tools#

• Cross-functional teams for systemic problems, where causes are not local and the fix needs management and several departments. Four to eight members, protected meeting time, and management agreeing in advance to act on realistic recommendations.
• Quality circles: permanent groups within one area tackling local problems. Andersen notes they never took root widely in the West.
• Concurrent engineering: a cross-functional team working in parallel instead of passing work along.

7. Planning implementation#

• AΔT analysis: A is the actual value of a measure, such as cycle time; T is its theoretical value once non-value-adding activities are removed; Δ = A ÷ T is the improvement potential. Set targets at or near T: ambitious but reachable.
• Tree diagram: break the plan into verb–noun actions, grouped into a hierarchy.
• Process decision program chart (PDPC): extend the tree by asking what could go wrong with each action and planning countermeasures. Use it for large, novel or high-risk changes.
• Force-field analysis: list the forces for and against the change, show their strength, and plan countermeasures for the strongest opposing forces.