I'd like to post some thoughts on the Siemens Healthcare Kanban implementation and case study. This case study from the United States has received a lot of coverage and the author and his Kanban coach have appeared at several Lean Kanban conferences presenting the story.
The story written by Bennet Valet of Siemens appeared on InfoQ and as such has caught the attention of many people. You can read it here...
Kanban at Scale – A Siemens Success Story
We've been reviewing this case study as a group exercise in KCP Masterclasses throughout this year. It is worthwhile posting some thoughts on this story because of both how it has been presented by Bennet Valet and Daniel Vacanti publicly and some of the claims made in the article.
Firstly, the good news, the story discusses the rollout of kanban systems across multiple services within a business unit at Siemens Healthcare in Malvern Pennsylvania. Around 500 people were involved and they already had Scrum implemented across product development. So this is a large scale Scrumban story where a business unit is adding Kanban to an existing Scrum implementation in order to take it to the next level. There are doubtless many companies interested in this scenario and the outcomes.
It seems that defect rates dropped dramatically after Kanban was implemented and productivity increased by approximately 35%. Achieving faster delivery from a group of 500 people by as much as 35% is a significant result and will have caught the attention of others in similar corporate situations.
So what are the problems with this case study and how it has been presented publicly? Firstly, there has been a salaciousness about the style of presentation and some of the claims made belong in the field of tabloid journalism. What is behind this behavior, I wouldn't like to speculate, but it has caused some chatter in the community that needs addressing. Secondly, there are some claims in the story stated as fact, and I have no doubt the author believes them to be facts but they don't stand up to scrutiny from reading the content of the case study.
(1) The claim is made that this Kanban implementation did not follow the advice and guidance taught in the KCP Masterclass and hence bucked the playbook used by professional Kanban coaches. This claim is not valid and needs refuting
(2) The claim is made that implementation was done in a revolutionary fashion and did not pursue the evolutionary approach of the Kanban Method. This is not valid and needs refuting.
(3) The claim is made that the initiative was run using Ackoff's Idealized Design approach and not the established approach of STATIK and the leading change method taught in the KCP masterclasses. While I truly believe the author believes this to be true, it does not stand up to scrutiny from the detail of the story.
(4) The claim is made that Kanban would not have worked were it not for the adoption of the Agile development techniques and prior adoption of Scrum. The implication being that people should not try Kanban without first doing the others. This claim is not substantiated by any detail in the story and appears to be purely speculation by the author.
Let's address each of these issues in turn...
First some background. We've reviewed this case study in 5 KCP Masterclasses in 2014. Several attendees have been current or former Siemens employees. We've reviewed it with many German speakers who are very familiar with Siemens as a business and its corporate culture. We've therefore been able to bring additional insight to the story that cannot be read by a member of the general public casually reviewing the story. Hence, we need to view this story through a lens that tells us there are 3 audiences for it; the Agile community within Siemens and those who have sponsored and paid for existing investment in Agile methods; Siemens as a wider corporate entity and its people who have certain cultural expectations; and the wider public audience interested in a large scale Kanban implementation in a major multinational.
(1) Was standard Kanban coaching guidance followed or not?
If we look carefully at what was actually implemented at Siemens, each kanban system isn't actually so different from the original Microsoft XIT case study from 2004-2006. However, the scale is much larger - about 15 times larger in terms of kanban systems and about 50 times larger in terms of people claimed to be involved. Though those who received Kanban training correlates with about 15 times larger.
So one way to view the Siemens Healthcare study is to consider it like Microsoft XIT Sustaining Engineering x 15. Given that the implementation was happening 8 years later and after considerable documentation of Kanban, considerable community development and many years of training, and developing Kanban coaches in the market, it is not unreasonable that someone would attempt 15 implementations simultaneously. We have seen many others of this scale around the world, including those by Brickell Key Award winners Richard Hensley and Amanda Varella, but none that have been so well documented.
(2) Was it a revolutionary transition or not?
Perhaps this claim demonstrates a misunderstanding by the author? Reading the story we could ask, did they start by understanding what they do now "team by team" (or system by system as we refer with Kanban)? and the answer is yes! Did they add kanban systems to their existing processes? again the answer is yes! Did they replace existing processes with new ones? No they did not! Did anyone get a new job title? Only those assigned to the "Flow Team" appointing as internal Kanban coaches.
Was the batch size large? i.e. more than 1 kanban system initially? Yes it was. In this respect, this does go against the conservative, cautionary, safe-to-fail approach typically taught in a KCP Masterclass. Typically, we look to start with one system and then spread the adoption in a viral fashion. We "flow" Kanban across an organization rather than dump it in a large batch. Does installing a large batch of kanban systems represent "revolution"? No it doesn't! And I believe this is the source of the authors misunderstanding of Kanban.
So it isn't a revolutionary transition, it's a larger batch introduction of an evolutionary approach.
(3) Was Ackoff's Idealized Design approach used?
... or was it in fact the STATIK method taught in the 2-day Foundation Kanban classes without calling it out in a branded fashion?
Here is a succinct guide to Idealized Design written by Russell Ackoff...
A Brief Guide to Interactive Planning and Idealized Design
Ackoff's method describes a collaborative small group approach to organizational redesign. We need to ask did Siemens actually redesign their organization? As described in the story, they did not. So while their facilitation style may have been inspired by that described by Ackoff, they clearly were not following the Ackoff approach. It is quite clear that they pursued "start with what you do now" and designed kanban systems for each service within their product development organization.
Those who've attended 2-day Foundation/Practitioner Level kanban training will know that the 2nd day contains a series of facilitated group exercises to design a kanban system by first asking, what do we do now? who are our customers? what do they ask us for? and what are their expectations? This is known as the STATIK (Systems Thinking Approach to Implementing Kanban) and was presented at the Boston conference in 2012. A conference attended by Bennet Valet, the author of the Siemens case study.
There are without doubt some similarities between STATIK and the facilitation and collaborative design style advocated by Ackoff. The differences between Idealized Design and the Kanban Method lie in organizational redesign versus "start with what you do now" and the leading change approach taught in the KCP masterclass. There is no evidence in the detail of the story that Ackoff's method was actually followed in reality. However, the story does closely match a larger scale implementation of what we did at Microsoft 10 years ago.
(4) Would Kanban have failed had Siemens not already adopted Agile development practices and Scrum?
This is clearly a speculative question and the author evidently has his opinion. Do we know that Kanban works when Agile practices or Scrum are not present? Yes, obviously we do! The first case study didn't involve anything Agile and most of the early copycat implementations such as Robert Bosch again did not involve any Agile practices or methods.
The riddle of this closing claim in the article is most certainly answered by the 3 audiences for the article. It was clearly important to signal to Siemens Agile adopters, advocates, and sponsors that their existing investment was valuable and shouldn't be discarded.
Meanwhile, the general theme of the article and public claims at conferences that Siemens pursued a revolutionary approach rather than an evolutionary approach is most likely explained in a similar way. Siemens is a very German company. As such they like to deliberate very carefully, but once they make their mind up to do something, they want to commit on a large scale and implement things quickly. An evolutionary approach goes against the Bavarian engineering ethic of "Design Teknik." It isn't just a Siemens cultural thing, it is a wider German engineering profession cultural norm. The article appears to be written in language that will appeal to that audience. Unfortunately this means that language that can be misinterpreted elsewhere is used prominently.
I actually think that the Siemens story is incredibly educational and I'd like to thank Bennett Valet and Daniel Vacanti for making it happen and being to publicly vocal about it. The whole community benefits from this. I very much consider the Siemens implementation as exemplary and it demonstrates that Kanban coaches are trained to adapt the rollout to the situation and the cultural and environmental aspects at the client. The Siemens story shows that implementations similar to Microsoft XIT but implemented on a business unit wide scale can have a truly remarkable impact on business performance.
Kanban coaches are taught to make the size of the J-curve as large as you can get away with without invoking internal resistance, then let the evolutionary process work from there. As such Siemens represents a larger J-curve than simply sticking a few boards on the wall and visualizing existing processes, but the J-curve is no more dramatic than we achieved at Microsoft when Dragos Dumitriu and I got all of this started.