Six Sigma arrived with an extraordinary claim: by applying statistical rigour to your processes, you could reduce defects to 3.4 per million opportunities. Motorola popularised it. General Electric made it famous. By the 2000s, every Fortune 500 company either had a Six Sigma programme or felt embarrassed about not having one.
The premise is sound. Define, Measure, Analyse, Improve, Control — five phases, each with specific tools, each grounded in data. Statistical thinking replaces gut feeling. Variation becomes the enemy. Control charts, capability studies, and hypothesis tests give you a language for quality that doesn't depend on who shouted loudest in the meeting.
But here is what happened in practice at thousands of companies. Six Sigma stopped being a problem-solving methodology and became a certification programme. The belts — Yellow, Green, Black, Master Black — were supposed to signify demonstrated competence. Instead, they became HR credentials. You attended a course, passed a multiple-choice exam, submitted a project, and received a title for your business card.
The organisation now had hundreds of Green Belts, dozens of Black Belts, and a handful of Master Black Belts. What it didn't have was fewer defects. The certification body — usually internal — had every incentive to pass people. A high failure rate would make the programme look bad and reduce enrollment, so standards drifted.
How the Belt Hierarchy Corrupted the Methodology
The belt hierarchy was borrowed from martial arts, which was either a brilliant marketing decision or an accidental confession. In martial arts, a black belt means you have mastered fundamentals through years of practice under direct supervision. In Six Sigma, a Black Belt often means you attended four weeks of training over six months and submitted a project report that someone reviewed for statistical correctness but not for business impact.
This created a predictable distortion. People pursued belts for career advancement, not for process improvement. The certification became the goal. The project was the vehicle for the certification, not the other way around. And since the training revenue depended on throughput, nobody had an incentive to fail candidates.
A Green Belt project was supposed to deliver measurable financial impact. In practice, the financial impact was often calculated using creative accounting. A process improvement that saved $50,000 in theory — based on assumptions about labour rates, defect costs, and throughput that nobody verified — was reported as $500,000 in annualised savings on the project charter. Finance signed off because that was the process, and the process had to be followed even when the process was the problem.
Black Belt projects had the same issue at a larger scale. A Master Black Belt — supposedly a mentor who had delivered millions in savings — was often someone who had been in the organisation long enough to accumulate project credits, not someone whose projects had actually survived contact with reality. The promotion criterion was the number of projects completed, not whether the improvements held twelve months later.
When Savings Claims Become the Defect
The Belt Programme Economics
Here is the expense report your organisation never filed: the fully loaded cost of every Black Belt pulled off their regular job for six months of training. The opportunity cost of every engineer who spent their days preparing DMAIC presentations instead of designing better products. The consulting fees paid to the firm that certified your Master Black Belts so they could certify your Black Belts so they could certify your Green Belts — a pyramid scheme of credentials.
A typical Black Belt programme costs between $15,000 and $30,000 per person in training and certification fees alone. For an organisation with 100 Black Belts — not unusual for a large manufacturer — that is $1.5 to $3 million in direct training costs. Add the lost productivity, the project hours, the internal infrastructure, and you are easily at $5 to $10 million per year.
Now ask the uncomfortable question: how many of those Black Belt projects delivered real, verifiable, sustained financial impact? Not the inflated numbers on the project charter. The actual, audited, twelve-months-later, did-the-process-actually-change impact. In most organisations, the honest answer is 10 to 20 percent. That is the same ratio you would get from almost any improvement initiative that didn't involve statistical software or coloured belts.
DMAIC as Data-Driven Theatre
The five-phase DMAIC framework is genuinely useful when applied honestly. Define the problem. Measure the current state. Analyse root causes. Improve the process. Control the new process to hold the gains. But in a belt-driven culture, each phase became a deliverable to check off rather than a discipline to practise.
Define became an exercise in scoping the project so narrowly that success was guaranteed. Instead of reducing warranty returns by 30%, the problem statement became reducing Dimension A variation on Part Number 12345 from 0.15mm to 0.08mm. This is a real problem, but it is a tiny problem. Solving fifty tiny problems while ignoring the systemic issue causing all of them is not improvement. It is whack-a-mole with statistical software.
Measure became a data collection exercise where the measurement system itself was never validated. People who had completed Gage R&R studies in training proceeded to collect data with uncalibrated instruments, inconsistent operator methods, and sampling plans that captured the wrong population. The DMAIC report included a Gage R&R slide because the template required it, but the actual data driving decisions came from whatever was easiest to measure.
Analyse became a parade of statistical techniques applied because the belt curriculum required demonstrating proficiency, not because the data called for them. ANOVA on a sample of twelve data points. Regression models with nine predictor variables and twenty-three observations. Multi-vari charts that looked impressive in a presentation but showed effects smaller than the measurement error. The statistics were technically present but substantively meaningless.
Improve became whatever change the team had already decided on before the project started. The statistical analysis existed to justify the predetermined solution, not to discover it. Control became a control plan that was filed in the same binder as the PFMEA and never looked at again. The process reverted to its old behaviour within months, but since the project was already closed and the savings already booked, nobody measured the backslide.
What the Shop Floor Already Knows
I have audited plants where the operators told me exactly which machine drifts on the afternoon shift and why. They had been telling supervisors for years. Nobody listened. Then someone with a Green Belt arrived and asked them to fill out a fishbone diagram. They wrote down what they had been saying all along. The Green Belt put it in a chart. The chart went into a binder. The binder went on a shelf next to the ISO 9001 procedures.
In the best case, operators will tell you about one Black Belt who actually came to the floor, watched the process, talked to people, and helped fix something real. That person exists in every organisation — usually one or two individuals who would have been effective regardless of the methodology because they care about the work and they respect the people doing it.
In the worst case, operators will describe the Black Belt who arrived with a laptop and a template, asked questions whose answers they didn't understand, collected data they never shared, ran analyses they couldn't explain, presented results that didn't match reality, and then left to work on the next project. The process didn't change. The defects didn't change. The only thing that changed was the colour of the belt on the person who briefly stood in their work area.
Six Sigma didn't fail because the statistics were wrong. It failed because the organisation didn't listen before the statistics and didn't change after them.
The belt system created a layer of intermediaries between the problem and the solution. Those intermediaries had more invested in their certification than in the outcome. The tools of Six Sigma are not the problem. Control charts are useful. Design of Experiments is powerful. Capability analysis is informative. The problem is what happens when the tools become the purpose rather than the means.
Recovering DMAIC From the Belt Industry
Certification Culture vs. Problem-Solving Culture
What certification-driven teams do
- Select projects that satisfy belt curriculum requirements
- Calculate savings on the charter before the process changes
- File the control plan and close the project within six months
- Promote based on number of projects completed and belts earned
What problem-solving teams do
- Select projects based on strategic priorities: warranty, scrap, OEE
- Audit savings claims with the same rigour as product inspection
- Return to the process at twelve and twenty-four months to verify
- Promote based on demonstrated, sustained process change
The organisations that get real value from Six Sigma share characteristics that have nothing to do with the colour of anyone's belt. Leadership involvement at the gemba matters more than the statistical sophistication of the reports. The executive who reviews project results by going to see the actual process change — not by reading the savings number on a slide — is the executive whose organisation actually improves.
Honest measurement is the dividing line. The organisation that audits its own savings claims with the same rigour it applies to its product inspections is the organisation that learns. Most don't, because the inflated numbers look better in the annual report and the deflated reality feels like failure. Sustainability tracking tells you the truth: twelve months after project closure, is the process still improved? If not, the control phase failed, which means the project failed.
Respect for the people doing the work is non-negotiable. The operator who has run the same machine for fifteen years knows more about that process than any Black Belt with a statistical toolkit. Six Sigma works when the belt is a facilitator who brings analytical tools to complement operational expertise, not an expert who brings templates to override it.
Strip the Hierarchy, Keep the Discipline
You do not need to abandon Six Sigma. You need to abandon what Six Sigma became. Strip away the belt hierarchy. Keep the tools. Train everyone in basic statistical thinking — not to certify them, but to equip them. Select projects based on what matters to the business. Measure results honestly. Track sustainability. Go to the floor and stay there until you understand the process better than the report does.
A Green Belt should not be a credential. It should be a skill. A Black Belt should not be a title. It should be a demonstrated ability to solve problems using data. A Master Black Belt should not be a rank. It should be someone who has personally led changes that held, who has mentored others to do the same, and whose track record can be verified not by a certificate but by the processes they touched.
The 3.4 defects per million is achievable. Some organisations have achieved it. But they didn't achieve it by certifying people. They achieved it by changing processes — one at a time, thoroughly, permanently — with people who cared more about the outcome than about the credential.
Your organisation probably has those people already. They are on the shop floor. They have been telling you what is wrong for years. Stop giving them belts and start giving them authority.