Making Sense of Deming - Commoncog

A comprehensive summary of W. Edwards Deming's ideas, whose System of Profound Knowledge is one of the most powerful things you'll find on the Operations side of the business expertise triad. Read this, so you don't have to read multiple books to apply his ideas.

This is a companion discussion topic for the original entry at

Thank you for this article. Superb! It has me thinking about my work deeply. In my work environment, I see the truth of this statement so often, quoting from the article:

Deming liked to say that mistakes are 85% a systems problem, and only 15% of the time the fault of the human.

So much of the analysis I see done is “after the fact”, not analysis of how training upstream from problems resolves issues.

I will have to read this article a few times. Thank you for the great writing.


One thought I had while reading the Deming piece that potentially explains why the companies that adopted Deming no longer exist but Deming’s ideas can also form Process Power that Toyota, Honda, and all other Japanese car manufacturers have:

Operational excellence doesn’t matter if the shape of your game is changing. If it doesn’t, it can eventually compound into Process Power that allows you meaningful advantage in your industry.

The companies listed that didn’t make it - American based and in small electronics and textiles industries, are in industries that globalization impacted as Asian economies developed. Governments of Taiwan, Japan, or Korea all targeted textiles and low end electronics manufacturing as industries to subsidize and spur economic growth. In this context, the shape of the game for these manufacturers was changing - no amount of operational excellence can compete with labor likely 1/10th the cost elsewhere in the world, and hence they slowly died out despite adopting Deming’s best practices.

In Toyota’s case, the shape of the game for cars hasn’t changed all that much. Outside of EV’s, high end sports car brands, or brands propped up by heavy government support, most major car brands that sell at scale were founded between 1900-1950 and compete under the same model. My guess is the high complexity of managing the auto supply chain, the scale required to enter this market, and the lack of a major technological leap in the auto industry (cars certainly are better than 50 years ago, but progress happened incrementally), make it hard for there a gap to exist that new entrants or a new business model can take advantage of. Under this comparatively stable environment, Toyota’s adoption of Deming’s philosophy in operations eventually formed Process Power, an entire system and philosophy of manufacturing that becomes hard to copy, and hence their advantage in the car industry.

What’s interesting is I think in the next 10 years the shape of the game for auto may be changing with EV adoption rising. New EV’s are competing on an axis that more resembles consumer electronics - faster release cycles, self driving, emphasizing software and hardware integration, direct to consumer sales etc.

Under this lens, a few questions off the top of my head

  • Does Toyota’s Process Power still hold if EV is the new game to play or if once again, we’ll see that operational excellence isn’t enough.
  • As an incumbent with existing Power, how do you find Power again once the market shifts? Would love a follow up from Helmer detailing cases about incumbents that successfully pivoted once the market shifted and found a route to continuing Power.
  • Does that mean in the Market, Ops, Capital triad, Market should always be the highest order bit? Until you are sure that the shape of the game isn’t changing, then you focus on supply and capital.

The example of failed Deming companies are clear examples of where “Doing things right” didn’t help “doing the right things”. SPC, variation, and Deming, to the limit of my knowledge of these areas, are of little help with making strategic choices and avoiding creative destruction when technology evolves. I think this is true for many frameworks born in the operations/supply side of the business expertise triad (TOC has the same issue).

With regards to goal setting and Deming, i had two spontaneous thoughts:

  1. Goal setting must be fine when we have a process that is not performing to the level of the voice of the customer. I.e the goal is to have a process that works in our context its fine to have a goal. The goal is to have stability at a level that is acceptable to the voice of the customer. (Perhaps this is obvious or nonsensical or both)
  2. Goal setting that is arbitrary and not connected to the system goal leads to disaster every time. This is clear from Deming, clear from Goldratt and most others. Which is why I find goal trees, metric trees etc to be a way to combine goal setting with overall system improvement. There are probably logical holes in this lien of reasoning, but I think it mitigates some of the issues Deming place on goal setting.

Wow, I had no idea I was a follower of Deming :slight_smile:

All of the things you find challenging about his philosophy are things that I found so intuitive that I had trouble understanding why people did things differently.

I always hated OKRs. If I knew exactly what I could produce, then I wasn’t near the edge - I was playing it safe. My only responsibility was to grow the business, and I didn’t know where that might take me during a quarter or a year. When I joined Google, I got incredibly frustrated working with other teams who refused to change what they were doing when I came to them with an ask.
Me: Why not?
Them: Because it’s not on my OKRs, and I am being evaluated based on my OKRs.
Me: But this is really important for the business! Let me show you the data!
Them: If it was important, it would be on my OKRs
Me: We just discovered this!
Them: Then my VP would tell me to change my OKRs.
Me: (sputtering in anger)

Similarly on the idea of individual recognition and compensation. I think I’m an unusual case in that I can be a superstar or I can be below average. I don’t think I change that much between jobs, so it’s more a function of how I fit with the system. I was often at my worst when I could see the system was broken, and leadership wasn’t doing anything about it - then I would work to actively undermine them, because they weren’t leading in the way I thought they should.

Side note: I once went and yelled at the Google VP of Compensation because I thought the whole idea of individual compensation undermined the principles of collaboration and teamwork that Google espoused. He didn’t agree, obviously. But I feel retroactively justified by Deming’s arguments :slight_smile:


Adding an anecdote I heard related to OKRs from a colleague.

Like many product-led growth (PLG) SaaS companies, there was a low friction self-serve option, and a significantly-pricier “contact us” enterprise bundle. The single sign-on (SSO) feature was only reachable through the enterprise bundle.

They had a lot of self-serve customers, currently paying <10k, who wanted SSO and were willing to pay for it. But jumping to the enterprise tier would have tripled the customer cost, which was unrealistic. There’s also a negative reputation in gating SSO (a security feature) behind significantly higher cost, the so-called “SSO Tax”.

To find a solution, this colleague did a bunch of interviewing/surveying and found that if they unbundled SSO and let customers buy it separately, the increased revenue should far outweigh any losses from the enterprise side. Great, so ship it?

Nope. They faced incredible internal drama from the sales team. The self-serve/mid-market team was happy with the proposed change. But the enterprise sales team was furious, since SSO is a big stick to close deals, and taking that away would make it much harder to hit their quotas - even though top-line revenue and customer happiness would increase, both good for the company.

So there was a multi-month process of re-structuring quotas to appease the enterprise sellers, and then another lengthy process to try to smooth the transition from self-serve to enterprise (mainly by shrinking the gap between the biggest self-serve contracts and the smallest-possible enterprise contracts). But as far as I heard, there was no particular change to sales goals that would prevent a similar issue in the future.