Lean validation: past, present and future
Since 1620 plus ça change
Hans Baumhardt | 27 February 2017
Validated learning is a foundational principle of the Lean startup philosophy, build-measure-learn feedback loops are the process. We have been working with these validation loops for over 25 years, starting in manufacturing then aviation operations followed by iterative software development and now technology startups. Plus ça change.
In 1991 an MIT study of the Toyota Production System was published in The Machine That Changed the World How Lean Production Revolutionized the Global Car Wars, which first introduced the term Lean production. Some of the principles will still be familiar today:
Kaizen (改善) the practice of continuous improvement to reduce waste and improve operational efficiency of repetitive activities.
Plan-Do-Check-Act (PDCA) a method of critical thinking & validation for continuous improvement.
In 2011 Eric Ries published The Lean Startup, a combination of Lean and Agile principles to help entrepreneurs manage innovation risk when creating new products, services and business models under conditions of extreme uncertainty. Some of the principles are eerily familiar:
Build-Measure-Learn (sequence may vary) from PDCA. Continuous learning and validation.
Lean from manufacturing. Reduce waste or burn rate to improve operational efficiency on repetitive activity, such as the iterative validation of a business model, market or product.
Agile from software development. Short time frame iterations with frequent customer involvement for fast outcome adjustments, such as the iterative validation of a business model, market or product.
Automotive production may appear to be the operational antithesis of feckless fail fast and pivot tech startups, but they both want to optimise repetitive processes to reduce waste.
Whilst the definition of done, from good enough through excellence to delight will vary due to the sunk capex of manufacturing production lines vs. ephemeral software, the foundation for successful outcomes are the same: discipline in critical thinking & validation.
The experimental method has been around for a thousand years (Alhazen ~1000CE), evolving more recently into statistical control feedback loops and the Lean business validation loops.
1620 Francis Bacon – investigative method: hypothesis – experiment – evaluation
1939 Walter A. Shewhart – statistical control cycle: Specification – Production – Inspection (or Plan, Do, Check – PDC). Developed at Bell Labs.
1950 W. Edwards Deming – wheel elaborated over time into: Plan – Do – Study – Act (PDSA). Lucky find document: A history of PDSA from 1610 to 1993
1985 Kaoru Ishikawa – Total quality control process: Plan – Do – Check – Act (PDCA) the heart of Toyota Lean manufacturing.
2008 Eric Reis – Lean Startup (the book was 2011): Build – Measure – Learn (push loop)
2010 Kent Beck – Startup Lessons Learned Conference 2010: Learn – Measure – Build (pull loop)
2012 Tristan Kromer – Build Measure Learn vs. Learn Measure Build: Hypothesis – Metric – Experiment
2015 Steve Blank – Why Build, Measure, Learn – isn’t just throwing things against the wall to see if they work: Hypotheses – Experiments – Tests – Insights
These perspectives evolving over time help to grok the fundamentals of validated learning iterations, but it is becoming overly elaborate. In the spirit of original Lean manufacturing Muda (無駄) we tend towards simplicity, using language that is easy to understand and accurate, over elite precision.
In this context the investigative method of 1620 still works (just add iterations), but the Lean startup movement has done a great job of building popular awareness for Build – Measure – Learn to a Pavolovian condition. Is that the sound of dog food ?
If you look for perfection you will never be satisfied
First published by the author on hjbconsulting.uk