Lesson 1: Pareto Analysis and Stratification — Using Data to Focus on the Vital Few Causes

Setting the Scene: When Everything Feels Urgent, Nothing Gets Fixed

Imagine you are a team leader at a mid-sized automotive components plant. Your OEE dashboard is flashing red across multiple lines. Quality defects are up, unplanned downtime is eating into your targets, and your team is running from one firefight to the next. Everyone is busy, yet the chronic losses refuse to budge. The real problem is not a lack of effort — it is a lack of focus. Without a structured, data-driven method to identify which losses matter most, your Kobetsu Kaizen projects risk being aimed at symptoms rather than root causes. This is precisely where Pareto Analysis and Stratification become indispensable tools in your analytical arsenal.

Learning Objectives

• Explain the purpose and logic of Pareto Analysis within the Kobetsu Kaizen problem-solving framework.
• Construct and interpret a Pareto diagram using operational loss data.
• Apply stratification techniques to disaggregate complex data and reveal hidden patterns.
• Select the vital few causes that deserve priority attention over the trivial many.
• Integrate Pareto Analysis into the broader PDCA cycle at the Kaizen Board.

The Logic Behind Pareto: Separating the Vital Few from the Trivial Many

The Pareto Principle — often summarised as the 80/20 rule — states that roughly 80% of consequences come from 20% of causes. In a manufacturing context, this means that a small number of failure modes, defect types, or equipment issues are typically responsible for the vast majority of your losses. Kobetsu Kaizen projects are specifically designed as expert-team projects targeting middle-sized to large-scale problems through detailed analysis. Without Pareto Analysis, even experienced teams tend to spread their effort thinly across all visible problems, achieving mediocre results everywhere instead of significant improvement where it counts.

In the Kobetsu Kaizen methodology, Pareto Analysis appears explicitly in the toolbox used at Step 2 (Problem Representation), Step 4 (Root Cause Analysis), and Step 6 (Checking Solutions) of the structured problem-solving board. This means Pareto is not a one-time exercise — it is a recurring analytical tool that guides your team from understanding the current situation all the way through to verifying that countermeasures have actually moved the needle on the dominant losses.

To build a Pareto diagram from operational data, follow these steps:

1. Collect and categorise your data. Use tally charts, OEE logs, or defect registers to count occurrences by category — for example, defect type, failure mode, machine, or shift.
2. Sort categories in descending order from highest frequency or impact to lowest.
3. Calculate cumulative percentages for each category and plot both the bar chart (individual contributions) and the cumulative line.
4. Identify the cut-off point — typically around 70–80% cumulative impact — to designate the vital few categories.
5. Speak with data. Present findings visually at the Kaizen Board so that the entire team can see, discuss, and align on priorities before moving deeper into root cause analysis.

The phrase “speak with data” is a cornerstone principle in this methodology. Gut feeling has its place in operational experience, but when it comes to committing team time and resources to a Kobetsu Kaizen project, data-backed prioritisation is non-negotiable.

Stratification: Cutting Through Complexity to Reveal What the Data Is Really Saying

A Pareto diagram built on aggregated data can sometimes be misleading. Suppose your top loss category is “mechanical failure,” accounting for 45% of all downtime. That looks like a clear priority — but which mechanical failures, on which machines, during which shifts, and under what conditions? Without breaking the data down further, your team may design a countermeasure aimed at a generalised problem that does not actually address the specific failure mechanism driving the majority of the loss.

Stratification is the technique of splitting your data into subgroups — by machine, product type, operator, time of day, raw material batch, or any other relevant variable — to expose patterns that aggregate figures conceal. When applied in combination with Pareto Analysis, stratification dramatically sharpens the focus of your root cause investigation. It also supports the N5W Analysis (No defect, 5W1H) approach embedded in the Kobetsu Kaizen Board by helping teams answer not just what is happening, but where, when, and under what specific conditions.

Practical stratification variables in a manufacturing environment include:

• By machine or line: Is the loss concentrated on one specific asset, or distributed equally?
• By shift or time period: Does the problem occur more frequently during night shifts or at the beginning of a production run?
• By product or SKU: Are certain product variants generating disproportionate defect rates?
• By operator or team: Is there a skills gap or knowledge transfer issue contributing to variability?
• By supplier batch or raw material lot: Are incoming material variations influencing the process output?

Each layer of stratification produces a new, more refined Pareto diagram — progressively narrowing the field until your team can point to a specific, measurable, and addressable root cause candidate. This iterative approach aligns directly with the PDCA cycle embedded in the Kaizen Board structure: you plan based on stratified data, act on focused countermeasures, check results with updated Pareto charts, and adjust accordingly.

Practical Example: Reducing Surface Defects at Meridian Plastics

Meridian Plastics, a fictitious injection moulding supplier to the consumer goods sector, was experiencing a defect rate of 6.8% — well above their target of 1.5%. The quality team had been addressing defects in a reactive, ad hoc fashion for months with no sustained improvement. The plant manager decided to launch a Kobetsu Kaizen project and anchor it firmly in data from the outset.

In Step 2 of the Kaizen Board, the team compiled three months of defect data from their tally charts and quality logs. A first-level Pareto diagram immediately revealed that three defect types — sink marks, weld lines, and short shots — accounted for 74% of all rejected parts. The remaining eleven defect categories were deprioritised for this project cycle.

The team then stratified the sink mark data — the single largest contributor at 38% — by machine. Remarkably, one moulding press (Machine 7) was responsible for 61% of all sink mark defects, despite producing only 22% of total output. A second stratification by product type showed that the problem was concentrated on a specific high-wall-thickness component. With this level of precision, the team directed their Why-Why Analysis and Ishikawa (fishbone) diagram at a clearly defined, data-confirmed problem statement — not a vague category.

The countermeasures introduced — adjusted cooling time, revised gate design, and a targeted PM inspection for Machine 7’s temperature control system — reduced sink marks on that component by 83% within six weeks. The updated Pareto diagram at Step 7 (Check the Solution) confirmed the shift in the loss distribution, and the team moved on to tackle weld lines as the new dominant contributor. This is the orientation toward zero that Kobetsu Kaizen projects are designed to achieve: measurable, sustained, and evidence-driven progress.

Key Takeaways

• Pareto Analysis is a decision-