Lean Enterprise - Summary

I've started writing down my notes of Lean Enterprise: How High Performance Organizations Innovate at Scale. I was thinking of posting this in one go but I've noticed it will take forever. Instead I'll keep it updated as I write.

My overall idea of the book aligns with this reviewer's feelings:

This is not a bad book. But I am a bit confused about who it is targeting and why the universally glowing reviews. This book is almost entirely a synopsis of other books with almost nothing in the way of synthesising them into a coherent whole or bringing new data or anecdotes to the table....

That being said, I wouldn't probably rate it with 2 stars only.

As a warning, these are my notes. That means I took the liberty of keeping short (or skip) parts I already knew.


What's an enterprise?

A complex, adaptive system composed of people who share a common purpose.

What's a complex adaptive system?

you're not ready yet, see Auftragstatik

The idea of common purpose is different than having a vision or a mission. Says the book, but it seems everyone has a different opinion on this one. What's important I assume is not to focus on maximising shareholder value since they create a bias towards short term results at the expense of longer term priorities. They also reduce innovation by focusing on tactical actions. John Kay's Obliquity provides detailed research and analysis on profit seeking paradox.

Part 1 - Orient

Chapter 1

The archvillain here is Frederick Winslow Taylor, the creator of Scientific Management. According to him:

  • Job of the management: Analyze the work, break it to discrete tasks.
  • Job of the workers: Execute those tasks. No need to understand more than how to do their particular.

The protagonist is TPS (Toyota Production System). More specifically Kaizen (continuous improvement) powered by alignment and autonomy. A parallel topic was studied in a tongue in cheek manner with Gervais Principle.

Taylorism makes workers into cogs in a machine. TPS instead requires workers to pursue mastery, gives them purpose and a level of autonomy.

Intrinsic motivators produce highest performance in 21st century tasks. Extrinsic motivators (stick or carrot) decrease performance. See Dan Pink's work (mastery, autonomy and purpose)

Ron Westrum divides enterprise cultures in 3:

  • Pathological: Run by fear and threat.
  • Bureaucratic: Protect departments.
  • Generative: Focus on the mission.

2014 State of DevOps Report: Key finding is companies with high performance ITs are twice likely to exceed their profitability, market share and productivity. (Q: I find it hard to isolate and build this correlation. Maybe companies who have a high performant ITs have in the first the right culture to allow it, thus explains other success in profitability, etc ...)


(Or Mission Command vs Command and Control)

Command and Control, people in charge make the plans and people on the ground execute them, are how armies run. 200 years ago!! But with the defeat of the Prussian Army by Napoleon, things changed. At least in armies. Companies are still trying to mimic the old fashioned way of running an army. (I know because I had an argument with a CEO who gave an example of armies and benefits of deep hierarchical structures. We had an interesting discussion with me starting by "You might think so but, you know, once Prussian Army was defeated by Napoleon ..." No need to tell he didn't like the direction it went :) )

Clausewitz concepts on On War:

Fog of war: The uncertainty we face as actors in a large and rapidly changing environment, with necessarily incomplete knowledge of the state of the system as a whole.

Friction: Prevents reality from behaving in an ideal way, or as planned.

Friction and Complex Adaptive Systems: The defining characteristic is that its behaviour at a global level cannot be understood by analyzing its components parts. Many properties or behaviours "emerge" from interactions between events and components at multiple levels.

Bungay (in The Art of Action) argues friction creates three gaps:

Friction gaps

* image taken from www.slideshare.net/erikschon/agile-leadership-experiments-for-results

One line summary of the chapter.

Tell what and why you want, leave the how part to your subordinates.

Chapter 2 - Manage the Dynamics of the Enterprise Portfolio

Diffusion of Innovations, Everett Rogers, talks about the cycle all successful techs and ideas progress. In terms of ubiquity, innovations which are competitive advantage in their early days become a commodity and base a support for another innovation. Different strategies and techniques must be employed in different parts of the lifecycle. This is captured by explore vs exploit dichotomy.

Exploring new opportunities and exploiting existing ones are fundamentally different. Management practices effective in one will be harmful in another. Startups that discover a successful business model and cross the chasm often find it hard to transition into the next stage: Executing and scaling in a growth market.


This part talks basically of Lean Startup book ideas, the Build, Measure, Learn loop or have a value hypothesis, decide what to measure, create an experiment, which is called Minimum Viable Product and see if you have a product market fit.

The Loop

The trick is to invest a minimum work in the loop. And the idea is you can use these techniques not only in startups but also in enterprise where the uncertainty is high. Legacy enterprise structures have hard time allowing this but as it is in most matters it's a question of trade-offs


Products / companies that have crossed the chasm are optimised to exploit their business model. Research from Microsoft reveals that planning driven feature development fails to create real value for business most of the time. (Online Experimentation at Microsoft)

Planning fallacy: Inclination of executives to make decisions based on delusional optimism rather than on a rational weighing of gains, losses and probabilities.

Combined with large batch death spiral (adding more features during the project once we gain more information leads to projects over time, over budget and most of the without producing any value. How do we save ourselves?

  1. Define, measure and manage outcomes rather than output.
  2. Manage for throughput rather than capacity or utilization.
  3. Ensure people are rewarded for favoring a long-view system level perspective over short-term goals.

HiPPO to rule them all

On a commissioned survey on 161 global business decision makers, 13% admit that the decision factor is HiPPO and 47%, almost the same, decision by committee.

3 Horizons Model

A sensible way to manage enterprise portfolio of products.

  • Horizon 1: 0 to 12 months, generate today's cash flow
  • Horizon 2: 12 to 36 months, today's revenue growth + tomorrow's cash flow
  • Horizon 3: 36 to 72 months, high risk, options on future high-growth

A couple of examples from companies' allocations on horizons:


Key takeaway: Cannibalizing your own business might be a good idea. (Amazon Kindle, Marketplace)

Part 2 - Explore

Chapter 3 - Model and Measure Investment Risk

The starting idea for the chapter is that we generally measure things which don't have information value and we don't measure things that do. (How to Measure Anything, Douglas Hubbard)

One of those metrics with no information value is development cost for a software project. Detailed planning and projections are generally produce waste since the most important unknown is whether the project will be cancelled. The second is its utilization. Focus on the fuzzy front end activities for radical process improvements (kaikaku)

Applying the scientific method

Basically, stop saying requirements and call them hypotheses. This implies two key differences:

  • Stop using detailed planning as a way to manage risk
  • Instead of creating only one plan, iterate by running series of experiments in order to discover a product/market fit

A common objection is that those measurements cannot fully represent a finished end product. That's based on a false understanding of measurement. It's not to gain certainty but to reduce uncertainty.

Expected Value of Information: Hubbard defines the value of information as follows: “Roughly put, the value of information equals the chance of being wrong times the cost of being wrong..."

OODA Loop - Observe, Orient, Decide & Act

John Boyd's model of disrupting your opponent and how humans interact with their environments. This is where Eric Ries borrowed his Build, Measure, Learn loop idea. There is a whole domain of psychology, military strategy and cultural aspect behind Boyd's ideas but I'll allow myself the naivety of summarizing the principle as operating inside opponent's OODA loop.


One common mistake is to interpret the loop as a linear flow starting from Observe ending in Act. The reality is you can perform all of these activities simultaneously and there are multiple feedback and feed forward loops between each of them.


comments powered by Disqus