I’m a huge fan of continuous improvement, the “lean enterprise” philosophy, and just making things more efficient. There are different things that draw people to software development and one of them is automating repeated and wasteful tasks in their lives. This is part of what Six Sigma is about applied to an entire organization. In this book review I won’t give all the content of the book but rather an overview of topics covered and how well I think it was communicated.
- This book goes over what Six Sigma is and how it is a data driven and criteria focused methodology for improvement
- The topic is a business and organizational topic but can be applied to any process including doing chores or planning a road trip
- There are a lot of relatable examples that can be used to teach the topic if you’d like to share these ideas with your team, friends, or family
- The second half of the book goes into statistics, data presentation, and data interpretation
- Measurement is the first step to improvement
- There are some good Lord Kelvin quotes about measurement referenced in the book
- Overview of the DMAIC process
I recommend this book to those who:
- Already have a Six Sigma leader in their organization and want to know what it’s about in a single reference
- Want a deeper understanding of which data visualizations are relevant and which are not
- Those who want an overview of the key components of Six Sigma (only read the first half)
- Six Sigma is about measuring inputs to your system, identifying variance, and controlling or reducing the variance in your process.
- This can be applied to any type of process including manufacturing, software development, or hiring.
- You must measure your all targets whether they are a quality bar (ex. physical measurement or latency) or the time it takes to produce an item.
- Both the quality and efficiency measurements are needed along with other factors you want to improve or maintain at a certain level.
- This is a practical and understandable use of statistical and experimental methods.
- Focus on eliminating special cases of variance or error rather than start with the common cases. Once you eliminate special cases then any improvements from common cases will be more impactful and less overshadowed by exception scenarios of variance.
- You need to measure all inputs and outputs along with your goal to figure out what your variance is and to see improvement over time.
- Try to find the factors that go into your process that have the greatest impact. Improving these first will provide you with more leverage to convince the organization to change.
- Follow the DMAIC process in order to achieve improvements.
- There are different types of improvements or change:
- Thinking for breakthrough – affecting cultural or thought change
- Processing for breakthrough – improving existing processes
- Designing for breakthrough – change or improve how new processes or products are created
- Managing for breakthrough – change leadership and hiring to create a culture of improvement
- Six Sigma makes everyone a leader for improvement who cuts through assumptions and fights for betterment.
- This type of organizational change needs to be driven from top down. You may have small scale success in an isolated team but you can’t do anything on a larger scale without management driving it.
- All aspects of an organization must be involved including non-production roles like finance or HR to define their expectations and results.
- A tool to turn an everyday problem into a logical plan and solution
- Problem definition
- Identify impacted processes
- Create a data driven definition for the problem
- Create a data driven solution goal
- Create a verification plan for the solution
- Design a practical solution based on the data driven goal
- Implement and see results!
- Typically improvements from this process should benefit all customers and stakeholders.
- There are different brainstorming techniques the book goes into that can be used with different groups of people to identify problem areas that feed into the definition phase.
Identifying the problem and target:
- Determine what needs to be improved (or the Y)
- Identify the impacted projects or processes
- Determine a baseline for Y of what is currently observed
- Quantify the cost and impact of this problem
- Write a problem statement: improve [metric] from [baseline] to [goal] in [time] with [impact] to [business goal]
- Identify key individuals and build a team: the best person to lead an improvement is the owner of the process.
- Obtain approvals from stakeholders and launch the project.
- If you have more than one Y to improve this may indicate your scope is too broad.
- Initial baselines may not be accurate but will improve with more measurements over time.
- Long term metrics are more valuable and relevant than short term or short-timespan metrics. Snapshots can be misleading and you’ll only identify trends of improvement over longer periods of time.
- Avoid biasing the project by putting parts of the solution into the problem statement.
- Create specific and measurable statements to convince management of the value and to ensure all parties impacted can understand the value.
- Identify the entitlement in your process: this is the best possible performance you can achieve by adjusting the inputs you can control.
- Work to identify hidden processes which likely represent areas for improvement.
This book went over various concepts in detail with examples of each. I will only list them out here:
- Mean, median, mode, and range
- Sum squared error
- Standard deviation
- Short term variance: “common” variation that is mostly random
- Short term standard deviation
- Long term variance persists beyond short segments and is typically caused by “special causes” (remember that is what we are trying to eliminate first)
- Short term variance should always be less than long term variance
- Poka-Yoke is a way to prevent variance in both special and common cases
At this point in the book they started to go into different ways of plotting the data, different types of visualizations, the pros and cons of each type of visualization, how to interpret the data from each visualization and I became very bored, very quickly. I stopped reading.