As part of the European Lean IT Summit, Institut Lean France organized a couple of master classes. One topic was the Obeya (by Sandrine Olivencia, Pierre Jannez and Dominique de Premorel), the other was on the A3 by Catherine Chabiron.
I was fortunate enough to attend the latter. This blog post about what I have learned while studying the very core of Lean thinking, the tool that allow to develop people before developing product, the formalized support of the scientific method of problem solving : the A3.
Warning : this post is only about the theory content of the training : it misses half of it which is the class studying a real case and working its way on the A3. In other word, reading this post will not train you : you need to do A3 and do the training to make sure you avoid common pitfalls …
We shall soon have a dedicated blog post to Catherine in the form of an interview. In the meantime, you only need to know that Catherine is Lean Office Manager at Faurecia (a worldwide top ten automotive equipment supplier) and she is a very respected figure in the Lean world.
Lean objective is the full customer satisfaction. In order to achieve this Lean relies on Standards and Kaizen (continuous improvement) as foundations, Jidoka (Providing machines and operators the ability to detect when an abnormal condition has occurred and immediately stop work) and Just-In-Time as pillars. Those pillars are a direct consequence of Toyota history. Jidoka because Toyota started doing loom machines : whenever there was an error, it was much better to stop the machine that keep it working and wasting fabrics. Just-in-time because when Toyota entered the car manufacturing market, they did not have the investment to compete with mass production system : just-in-time was just a way to relieve the company from the financial burden of stocks.
William Edwards Deming
Deming has been a massive inspiration for the original Lean Movement. His concept of profound knowledge has been directly inspiring Lean thinking. Continuous improvement, training on the job (as opposed to in training sessions), focus on quality insurance as opposed to quality final control (i.e. built-in quality), breaking down silos across functions, focus on data (Deming was a statistician with this glorious quote : In God we trust, all other must be bring data) etc … all these basic principles have fully permeated the Toyota culture.
Deming also is considered the father of PDCA (Plan Do Check Act) or Deming wheel, which he is not really as the concept is from Walter Shewart another statistician he admired.
Data, Information, Knowledge, Understanding and Wisdom
Catherine draws an interesting parrallel between PDCA and the DIKW framework (#hypertextual has already discussed) customised to DIKUW with understanding added into the mix. Even if this model has fierce opponents such as Dave Snowden, I do like its simplicity :
- Data : facts on events or objects, raw data without specific meaning
- Information : Processed data answering questions Who, What, When, Where, How Many
- Knowledge : information put in its context, as consequence, connection, conversation. Answer the How ? question
- understanding : answer the Why ?
- Wisdom : Ability to see the long-term consequences of our action
The scientific approach
I believe the scientific approach goes as far as the Royal Society and Francis Bacon (great history told in Cognitive Surplus by Clay Shirky). Principles :
- Observe an effect (a gap Vs expectations)
- Formulate a theory
- Test and measure
- Identify anomalies Vs theory (the learning part is in the delta between the two as Hakan Forss reminded us at the #LeanIT2013)
- Formulate another theory
- Test and Measure
- Confirm and publish the theory (standard)
- Rework (standard) theory as new facts emerge
The PDCA works along these principles with “gap Vs expectation” replacing effect and “Standard” replacing Theory.
Here is the key of problem solving in lean : we start with a problem which can be measured as a difference with expectation and the end of the loop is a new or modified standard. To reach that phase we have been testing and measuring hypothesis (or counter-measures). We make sure we fully understand and describe the problem (remember the quote by Albert Einstein : “If I had an hour to solve a problem I’d spend 55 minutes thinking about the problem and 5 minutes thinking about solutions”) and we make sure we don’t jump to solution.
In other word, problem solving = spot deviation to standards (in a view to return to standard) Or absence or inadequate standard (which requires improving existing or defining a new standard).
Standards and Managers
In a Lean organization, the role of managers in this problem solving approach is to :
- develop and control the execution of standards
- develop the capacity of each team member to solve problems.
PDCA Structure
The PDCA can be breaking down into the following actions :
- Plan : Collect data and facts, reduce the scope by eliminating possible causes and define a corrective action on the confirmed cause
- Do : implement the identified action
- Check : Measure the result
- Act : Standardize the action
A3 Vs Paper board
The question is : when should we do A3 ? Should we do it for any problem ? Answer : you should only do A3 when the problem affects different parts of the organisation or different entities along the value stream. If problem fully resides in your team’s area of influence you just use a local paper board :
In any case, the approach still remains the same : you need to collect facts and data to make sure you protect your thinking process from your preconceptions (always ask yourself when you have a “solution” : how do you know ?)
A3 Structure
A3 format is for making sure you remain synthetic (A4 was too small). It is strongly recommended to use graphs and flows rather than text to save space and to improve the efficiency of the communicated message (errm … not like this blog post). Hint : replace adjective and adverbs with facts (and data when possible).
Reading a A3 is like a reading a story. A good way to see problem solving is as a crime investigation.
1. Header
The Title Defines what we want to do in the process to return to standard. The Leader is the person in charge of leading the investigation. She has a higher view. Most importantly she is the one learning while investigating the problem, the person that we develop. A good candidate might be the person at the end of the process or a person that we want to develop.
Remember the A3 also is coaching tool, the coach using the A3 to exchange with the Leader. The coaching will ask
2. Background
The aim of this section is to answer the following question : Why do we want to solve this problem ? Why is it important ? What is at stake ? What is it going to happen if we don’t ? We really need business data here : amounts at stake, risk of doing nothing etc …
3. Grasp Current condition
The aim is to ensure we are correctly describing the problem and not a possible solution. The idea is to understand the situation as of today. This is where the Leader needs to go and see for herself, to observe, to collect facts rather than opinions. The tools here is the 5W2H : Who / What / When / Where / Why / How / How Many, the tally counts, volume figures, Paretos.
The Is / Is Not question is a great way to eliminate causes and reduce scope. The idea is to bring it down to very operational and measurable issues.
From this analysis the idea is to identify the key points on which we focus (80/20 from Pareto, etc …).
Fishbone diagram (aka Ishikawa) can also be used along 5 major topics (4 M’s) : Manpower, Materials, Machines, Method, Environment. Catherine recommends to be careful as she expressed her doubts on the usage of Ishikawa as it can lead to many potential factors, instead of reducing the scope via observation and factor elimination, to a limited nb of probable factors.
4. Set Targets
The target has to be measurable (if not, how will you check that your corrective actions are effective ?), with a clear deadline, and we must remember to regular re-focus on that target
- What is the target, the one we have deviated from ?
- By when should we reach it ?
- How do we measure it ?
5. Analyse causes
The key concern here is to make sure we don’t listen to our preconceptions and don’t jump to conclusions nor solutions
The idea is to think about possible causes using the following causes :
- Find experimentation to confirm hypothesis
- Root cause analysis (see the great paper by Henrik Kniberg on his cause effect tool)
- 5 Whys (warning : avoid « dry » whys : make sure you put the actual verb). We are at the Understanding stage of the DIKUW model.
- Ishikawa diagram.
Think of this as a crime investigation. Again, use Is / Is Not questions and associated data collection to reduce the scope to a limited number of potential causes. Then the following table can be used to confirm / eliminate those potential causes on your A3.
| Topic | Observed effect (fact) | Possible cause | Investigation underway to confirm the cause | Confirmed ? |
6. Apply countermeasures
This is DO in the PDCA. Define the action list make sure each action has a owner and date.
|
What |
Who |
When |
Status |
7. Confirm results
This is the CHECK in the PDCA.
These are the questions we need to ask ourselves : Are the results confirmed ? Have we re-iterated the A3 if not ? How can we make sure our corrective actions really are mitigating the problem ? It is critical to measure the results of the outputs. You may also want measure the inputs though it’s not mandatory.
A very interesting point here : the frequency of measure drives the capacity to improve and modifying if problem is found. If you measure everyday, your capacity of investigation is 30 times more powerful than if you measure every month (cf introduction of innovator’s dilemma). One of the reason why in Scrum, shortest iteration (2 weeks) are much better : they provide more opportunity to learn.
8. Standardize
This is the ACT stage of the PDCA. This aims to answer the following questions :
- How are we going to make sure we keep the same level of performance ?
- What would we do to ensure we maintain the new performance level ?
- Who else, what other process can benefit from the experience ?
A standard is supposed to change and evolve on a permanent basis. If it doesn’t, it’s a rule or a procedure.
Standards are a key element of the Lean system. They can be defined as the best known way today to do a task. It can be a simple one page work instruction, or a check-list, with built-in quality controls and performance expectation (time, quality). They enable :
- Immediate detection of any variation (Vs standard)
- Reaction to deviation to return to standard
- Make subsequent problems visible – reveal them
We are at the Wisdom stage of the DIKUW model : this is how we formalise and share the learning.
The main thing now is to find your own problem you want to track down using a A3. Most importantly, you need to have someone experienced to coach you along the way. This really is where the Lean journey starts.
A3 Strategy and A3 Proposal
Beyond A3 as a support to problem solving, this tool can also be used for strategy definition. This is how it is used for Hoshin Kanri, refer to the excellent presentations by Pierre Masai (CIO Toyota Motor Europe) or Cesar Gon (CEO CI&T) at Lean IT Summit 2012 on that topic.
A3 can also be used to structure a proposal.
For these two specific A3s, we haven’t done much exercises so I don’t really feel comfortable describing those. They surely would require a dedicated supplementary day.
#hypertextual 
2 Comments