Skip to content

Analysis · Structure and leadership

Has Lean Six Sigma gone out of style?

In the US, Japan and Brazil, the central continuous improvement team is losing ground. The method lives on, now in the hands of the people who run operations.

By 2025, nearly half of all annual reports mentioned AI. Six Sigma showed up in 1.1%.

10-K filings with the SEC that mention each term, as a share of the year's total

Artificial intelligence

20100.7%
202549.7%

Six Sigma

20101.8%
20251.1%

Lean manufacturing

20062.1%
20251.4%
Source: SEC full-text search (EDGAR), author's count, October 2026.

What changed

Every public company in the US files an annual report with the SEC, the 10-K. I used the SEC’s full-text search to count, year by year, how many of those reports mention each term, and divided by the total number of 10-Ks filed that year. In 2010, 1.8% of reports mentioned Six Sigma. By 2025, it was 1.1%. Lean manufacturing fell from 2.1% in 2006 to 1.4%. Over the same period, artificial intelligence went from 0.7% of reports in 2010 to 49.7%.

The job market shows the same shift. Avery Point Group, a recruiting firm, has analyzed job postings in this field since 2005. In its 2022 study, postings asking for Six Sigma alone made up 13%, down from 27% in 2013. Postings asking for Lean alone rose from 41% to 62%.

Among operational excellence professionals, PEX Network’s 2024 survey of 184 respondents found Lean in use at 35% of companies and AI at 31%. Six Sigma didn’t make the top ten methodologies. In the 2026/27 edition, 61% say AI has become the most-used methodology in transformation work.

In Japan, the shift started earlier. Quality control circles are groups of production employees who meet to solve problems in their own area. In 1988, JUSE, the organization that created them, estimated roughly 750,000 circles and 5.5 million participants. By 2014, there were 45,438 registered circles. The two counts don’t measure exactly the same thing, but the gap between them is wide. The number of improvement suggestions per employee, which the Japan HR Association has tracked since 1960, has been falling since the asset bubble burst in the early 1990s. The average gain from each implemented suggestion, on the other hand, has gone up: fewer ideas come in, but they’re bigger.

Why it changed

3M adopted GE’s Six Sigma in 2001 under CEO James McNerney. The company used to draw one-third of its sales from products launched in the previous five years. By 2007, that share had dropped to one-quarter, and McNerney’s successor, George Buckley, pulled the method out of R&D.

At GE, Six Sigma lost momentum after Jack Welch retired in 2001. Larry Culp, CEO since 2018, led the company’s turnaround with Lean, with a different design. Plants run kaizen events almost every week with their own workers and managers, and corporate headquarters shrank from 18,000 people to a couple hundred.

In Japan, in November 2007, a court ruled that a Toyota employee had died from overwork and counted the hours he spent on after-hours quality activities as work. In June 2008, Toyota began paying overtime for those meetings. The “voluntary” after-hours improvement work that kept the circles going became more expensive and harder to defend legally.

The money moved, too. In PEX’s 2024 survey, AI topped the list of planned investments for the following 12 months, cited by 47% of respondents.

Who is affected

Owners who have built, or are thinking about building, an operational excellence function staffed with analysts and Black Belts. And the professionals in that function. The jobs are still there, but they increasingly call for Lean, data and automation, and less often for Six Sigma certification.

Where Brazil stands

Brazil has no public data series that measures this decline. What you can see is how large companies organize improvement today, and the pattern looks like the US one: the management system stays, but inside operations.

Vale measures the maturity of VPS, its production system. The score rose from 2.57 to 2.70 in 2023, against a target of 3.0, according to its Management Report. Embraer has run P3E, its Lean-based program, since 2007. In 2022, it signed an agreement with Toyota do Brasil to bring Toyota Production System specialists into its São José dos Campos plant. Ambev runs VPO, its plant management system, with machine-level data dashboards.

Gerdau shows the change in emphasis. Its 2024 annual report features an AI Innovation Center created that year and doesn’t mention Lean, Six Sigma or kaizen. In the report the company files with the SEC, Six Sigma appears as one of the tools its specialists use in product development.

Falconi, Brazil’s largest management consulting firm, invested R$100 million in AI and technology between 2022 and 2026. Sales in that practice grew 27.5% from 2024 to 2025, and the firm now also serves mid-sized companies.

What it changes for the owner

Continuous improvement moves off the org chart and into the routine. The department head owns the scrap, lead time or rework metric, in a short, frequent meeting with the number on the table. A central project team runs the risk of becoming a department that produces savings reports no one checks a year later.

Black Belts are still useful for one kind of problem: variation that calls for statistics, such as a process running out of spec or a defect that shows up in 2% of batches with no clear cause. For that, it makes sense to hire a specialist or run a project with a start and an end date. I wouldn’t hire someone to “own improvement” on a permanent basis.

AI fits the same logic. It speeds up the analysis, but someone still has to decide who owns each problem.

What I would look at first

I’d ask two questions inside the company. Does every key metric have a named owner, and how often is it reviewed? And of the improvement projects from the last two years, how many are still in place, with the gains measured? If the answer is “few,” what’s missing is the routine that sustains projects, and launching more projects won’t fix that.

What we don’t know yet

It’s too early to tell whether AI will deliver the gains improvement programs promised or repeat their cycle: announcement, enthusiasm, dedicated team and, a few years later, a quiet wind-down. The surveys show the intent to invest, but they don’t show results yet.

In Brazil, the data is missing. No public survey tracks improvement roles or programs over time, and the large companies cited here don’t represent the average.

Methodology note: the 10-K count uses the SEC’s full-text search, which counts documents containing the exact term. Total filings per year come from the EDGAR index.