Kanban on Track - Evolutionary Change Management at the Swiss Railways
Swiss Railways (Schweizerische Bundesbahnen, SBB) employed Kanban to transform the interdisciplinary technology department "Unified Network Objects" (UNO) from disappointing performance to predictable efficiency through a series of incremental improvements. The evolutionary nature of Kanban gained traction with early quick wins and resulted in better management and greater responsiveness to change. This is a brief report of a two year journey by the UNO department manager (Mike) and one of the external coaches (Sigi) who supported this journey in various ways.
The SBB is the biggest travel and transport company of Switzerland with 30 000 employees from about 80 different nations and with more than 150 different professions. Transporting 350 million passengers as well as 50 million tons of goods a year, the SBB operates the world's most intensively used track network.
The Unified Network Objects (UNO) department contributes to this efficiency by converting a large amount of raw data into an information network for railway schedule and control. It comes as no surprise that completeness, quality and consistency are key success criteria for this Information network.
Internally, UNO is an interdisciplinary department that builds on the expertise of IT people as well as physicists, mathematicians, train engineers and dispatchers in four core teams (analysis, development, test and data management). Along with the department manager, the release manager and the product manager the leaders of these core teams build the leadership team that shares responsibility for UNO´s success.
Externally, UNO is only a small part of a broad network of internal suppliers and clients (see Figure 1). Accordingly, it is both kind of a hub and a classical bottleneck with three mission-critical issues:
- the sheer amount of complex data;
- the synchronicity of data processing, maintenance and new requirements;
- the difficulty of reliable planning in a quickly changing environment.
(Click on the image to enlarge it)
Figure 1: The UNO network
Where UNO started from
At the beginning of 2012, UNO was facing some serious problems: due date delivery problems, quality problems, the stress of constant re-planning and an alarming number of hot fixes. UNO´s daily business was defined by task-switching and the chronic overload of many employees. Despite a high level of commitment in terms of a high level of identification with the product and long workdays the results were rather disappointing and frustrating.
But how to break out of this vicious circle of overload, stress and lacking effectiveness? How to achieve both a higher level of productivity and stress relief? On their search for helpful answers, the UNO leaders stumbled over the evolutionary change management, well known as the Kanban approach. The decision for Kanban was mainly driven by the basic assumption that there was the least risk of making the bad situation worse through the additional effort of a big change. The initial idea was an adoption of Scrum, but accidentally learning the principles of Kanban, the leadership team was attracted by:
- the incremental as well as respectful nature of the approach (“start with what you do now“);
- focusing on systemic improvements, first and foremost on the amount of parallel tasks and the work flow beyond team boundaries;
- pull-driven small steps rather than big change plans ;
- the promise of quick wins by applying easy to understand practices such as visualization, WIP limits and explicit policies;
- the importance of leadership at all levels in order to create a culture of continuous improvement.
How Kanban was introduced
The first Kanban artefact had been created in March 2012 at the first UNO leadership retreat with all team leads organized by the department manager. It was a huge, 2x4 meters cloud of stickies which showed all current work items and process steps of the UNO workflow. As expected, this cloud created transparency and was jointly considered the first big step forward. The cloud made obvious that there was simply too much work in the system. It helped to think about a different trade-off between demand and capability and even allowed UNO to get rid of a lot of things that had not been adding any value in the past. It also shifted the focus from specialized teams to the whole system supposed to deliver unified output. In hindsight, the leadership retreat provided not just an innovative format for a joint helicopter view and the visual management of what had been discovered. It also allowed the leaders to gain more common ground in terms of focus and flow. Last but not least it was a welcome opportunity for team building.
(Click on the image to enlarge it)
Figure 2: First draft of “Realisation Board” in April 2012
Encouraged by these positive results from the leadership retreat the next improvement steps were taken over the next couple of months:
- In May, the rather chaotic cloud was redesigned to reduce complexity and better fit the organisational structure. This resulted in additional boards for each core team and the key account management (KAM). Figure 2 shows the first design of the so-called “realisation board” with a first draft of work items (Post-its), blockages (small pink Post-its), process steps (development, test and delivery, each with two columns of “in progress” and “done”), swim lanes (which ones?) and classes of service (yellow Post-its for standard work, blue ones for bugs, pink ones for expedite).
- In June, a new meeting structure was introduced with clear goals, cadence and duration.
- In July, the physical tickets on the boards were connected with the electronic JIRA system.
- In September; regular releases were planned for the future.
A few weeks later, the two authors of this article met each other for the very first time at the Kanban Change Leadership Training, held by Klaus Leopold and Sigi Kaltenecker. Despite all the improvement steps that had already been taken, Mike Beyer raised some serious concerns: Had the evolutionary change really touched the ground? Had the right things been changed? Was the department truly on the road to a culture of continuous improvement? On further inspection over the course of the training a whole bunch of new challenges popped up.
First and foremost, the critical view on change and leadership aspects showed the downsides of UNO´s hands-on approach. First of all, important stakeholders such as requestors, data suppliers and clients had not been involved yet, because they were not seen as being affected by the evolutionary change. Secondly, the team was not well prepared for the Kanban change, since there was neither enough information about the change process nor a clear understanding of the evolutionary approach. Thirdly, despite the promising kick-off in the course of the retreat, the leadership team building got somehow stuck and was overwhelmed by the ups and downs of daily business Finally, it became crystal clear that the change efforts had been too much top-down so far. This resulted in a lack of decentralisation and shared responsibility. Thus, the Kanban principle of “leadership at all levels” became the signpost for further improvement.
From Kanban to Kaizen
Following up on the Kanban Change Leadership class, the two trainers Sigi and Klaus were invited to support UNO onsite at the SBB in Berne. This invitation was driven by the conviction that techniques and tools are not enough to establish a mindset of continuous improvement (Japanese: Kaizen).
The concept of kaizen comes from the Japanese and literally means “change for the better”. Kaizen means far more than just technical change management; it is a philosophy for life and work at the center of which lies the quest for constant improvement. Orientating all organisational processes towards customer value is thus just as critical as cultivating the right attitude. Process optimization goes hand in hand with a humanization fed by professional leadership, which always begins with the individual and never ends. Leadership values such as identifying opportunity, positive thinking, self-responsibility, and a solution orientation are an ever-present challenge.
quoted from Leopold/Kaltenecker 2014: Kanban in IT
In order to create a sustainable culture of continuous improvement, the technical system has to be complemented with social dimensions such as organisational dynamics, the psychology of change and team work (Leopold/Kaltenecker 2014).
Fig 3.1 Joint retrospective methods - Research groups on most important insights
Fig 3.2 Joint retrospective methods - Timeline
Fig 3.3 Joint retrospective methods - World Café Method
Two main topics were on the agenda for Sigi´s & Klaus´ visit to Berne: firstly, an intense one-on-one sparring between Sigi and Mike to review his point of view on the current situation; and secondly, the joint facilitation of a department retrospective to learn more about each team member's perspective. Since they were the first joint retrospectives within UNO, both interventions focused on the basic questions of “what went well” and “what could be improved” (see “Getting Value out of Agile Retrospectives” by Luis Gonçalves and Ben Linders 2013). After a brief introduction round, we started the joint retrospective with a so-called “Timeline”. Each team member was invited to write the most important events on pin-cards and position these cards along two axes: time (“when did it happen?”) and experience (“how did I feel about it?”). This allowed us to combine hard facts and feelings and resulted in a kind of a fever curve that was examined by small cross-functional groups. Since the timeline and all the cards were positioned on the floor, these so-called research groups symbolically went into their own history to apply a helicopter view and detect positive as well as negative phenomena (see Figure 3). Here is a list of the most important positive things that had been achieved so far:
- Better planning
- smaller batch sizes
- better testing
- better release planning
As to the most important areas to improve on, the group brought up the following topics:
- synchronization of the boards
- clear work flow across all teams
- further elimination of task-switching
- further clarification of policies
- general understanding of Kanban and the opportunities of evolutionary change management
These improvement areas were further discussed within a World Café set-up. This set-up helped to focus the conversation in a solution-oriented manner. After defining the hosts for each topic, we ran three rounds of 20 minutes conversations with different guests on each topic-table, applying the law of the two feet to allow for a free floating of interests. Afterwards the outcome, the well-known World Café „table cloths“ with everyone's written or drawn contributions on it, was briefly presented by the hosts before we started a plenary round on conclusions and commitments. At the end of the day, we got a pragmatic list of „Next steps“ to be implemented in various areas throughout UNO.
What did we learn during the first UNO retrospective? On the one hand, we discovered core problems and explored new ways to deal with them. The joint retrospective had been highly appreciated by all attendees and became the number one format to coordinate future improvement. Even more important, the department manager and the leadership team achieved a new level of clarity about the current situation. They started to see the wood for the trees. Building on the observed weaknesses and deficits within the whole team, leaders and external consultants were able to design a tailored coaching and training program, addressing the main issues for 2013. This program provided a mixture of technical skills training (“What is Kanban?”), leadership training (“What is evolutionary change management?”) and personal as well as team coaching (“How can I/we improve my/our capabilities?”).
(Click on the image to enlarge it)
Figure 4: UNO workflow, Kanban boards, and meetings
Encouraged by the balanced input, UNO developed a new “big picture” as illustrated in Figure 4. The whole workflow evolved from the interaction of Kanban board structures and organisational functions. Although each board had its own specific concern, everybody felt a strong need for clear overall responsibilities and communication channels. At a certain point, the UNO leaders created an overview of all necessary feedback loops and identified some communication gaps to be filled. Accordingly, the existing meetings and roles were adopted. From now on both cadence and duration of the meetings were managed by the meeting members.
In just four on-site days in total plus a three day Kanban Change Leadership class for the team leaders, the coaching and training program generated a lot of Kaizen impulses. At the end of the year, we were able to review many positive results. Here is a brief overview of the most important improvements besides the workflow and meeting structure:
- a better understanding of Kanban principles and practices on a broad level – evolutionary change management had become everybody's business and a key driver of the UNO overall performance;
- joint policies for a consistent workflow from the KAM board via the UNO board to the different team boards vice versa (see Figure 3 – blue boxes and red arrows);
- the introduction of a joint change board with a clear change flow, change-in-progress limits and explicit policies to better monitor the rather overwhelming number of initiatives throughout UNO;
- the introduction of service classes throughout UNO for better risk management using the cost of delay model;
- a stronger leadership team due to better understanding of the whole system, intensified collaboration and more trust;
- the continuous improvement of collaboration, meeting facilitation and board designs (see Figure 4, “Realisation Board” in January 2014 with a better visualisation including avatars, columns for development in progress, review and done as well as test management, test data management and test done, swim lanes for Definitions of Done, expedites, bugs and standard work and policies for the stand ups and the facilitation)
- presentations within the SBB, at community events and at Lean/Agile conferences;
- a broader sharing of knowledge and lessons learned through on-site inspections by other departments, the CIO and even colleagues from other companies. Leaders as well as team members showed their boards, explained their system and ran Q & A sessions with their guests;
- the acknowledgment of Kanban in the SBB project management methodology catalogue. This official approval reflected the value of the evolutionary change management approach and encouraged even more Kanban implementations throughout the SBB. Thereby, UNO became an often cited reference point for a new kind of change and the focus of a growing number of visits that made UNO employees start joking about “our Kanban zoo“...
Figure 5: “Realisation Board” in January 2014
Reviewing more than two years of Kanban in operation, the cultural change at UNO is obvious. By operating a continuously improving system, employees became more and more active change agents. Rather than just being passively affected by this system, they used it to identify problems and develop customised solutions. Most importantly, they created their own systems that enabled them to see not just their respective area of responsibility but the big picture as well.
Summarizing the impact of Kanban at UNO, there are four major aspects to be pointed out:
- Quick wins. As promised, it was easy to achieve positive short-term results, as triggered by the very first leadership retreat and the onsite follow-ups. Applying the low-threshold approach of Kanban, transparency was the first big step forward towards a different way of running the daily business.
- Better management. The transparency of work and processes enabled better communication and control at all levels of UNO. Besides, the popular principle of “stop starting, start finishing” led to a remarkable improvement of throughput as well as predictability and quality.
- Agility in action. Both visibility and intensified communication increased the level of attention to the ever changing environment. At the same time it also decreased the response time to these changes. All of a sudden, it was easy to anticipate new demands and overcome the defensive mode of being either too late or too slow.
- Cultural change. In the course of Kanban operation, the behaviour as well as the mindset of the UNO team members changed significantly. While leaders gradually overcame the command-and-control paradigm to learn a support-and-coach attitude, employees realised that it was on them to take the full responsibility for performance and decision making. After a relative short period of uncertainty, the opportunity for self-control was embraced with more and more enthusiasm. The boards became true team boards that were continuously used and improved. An increasing amount of „nonsense“ such as funny avatars or weird awards reflected the relaxed yet professional manner in which Kanban was applied. Despite this team autonomy, the transparency of both work flow and feedback, nevertheless, helped to stay aligned within a joint departmental design (see Figure 4).
Despite all the positive results , the success of Kanban at UNO was not a given. In retrospect, some traps and obstacles became clearer too. Perhaps the biggest trap has to do with the way Kanban had been introduced. Since Kanban seemed an easy-to-understand, easy-to-implement approach, some key lessons of change management had been neglected from the very beginning. As already mentioned, neither the most important stakeholders had been suitably involved nor a clear problem statement shared with the core teams. On the one hand, it was a classical top-down driven project that pushed people in a specific direction. On the other hand, Kaizen was thought to be a no-brainer, kind of something that automatically happens once Kanban is introduced. Certainly, this belief resonates with the “natural” tendency of IT professionals to reduce change to a matter of technical expertise. Thus, organisation was still seen as a machine that needs to be fixed by the right mechanics.
As the UNO case shows, Kanban does not necessarily lead to a culture of continuous improvement. On the contrary, after a few month the initiative got stuck in various ways. Some key players showed significant scepticism, the core teams were reluctant to use Kanban for their daily business and external stakeholders did not understand what was in it for them. For some, Kanban appeared to be the classical flash in the pan, kind of the next fashion wave of change to roll over the department. Regressive behaviour such as ignoring policies, working beyond WIP limits or neglecting blockades was encouraged, the old patterns of how we do things around here still a powerful resource. In short: Culture was about to eat Kanban for breakfast.
How did UNO manage to get out of these traps? As it turned out, one of the root causes of many problems was the total absence of retrospectives or any other kind of meta-reflection. For sure, there was also a lack of understanding and agreement, two of the key ingredients of Kanban´s survivability agenda according to Mike Burrows. But since there was no culture of inspection and adaption established, people continued to be rather blind to what they suffered from. As often, regular time-outs weren't seen as valuable in the beginning of UNO´s Kanban implementation. Besides, there was not enough trust in people's capability to solve their own problems. For sure, a certain amount of fear that once everybody speaks openly managers might lose control and teams end up in chaos was involved too.
Of course, retrospectives are not a silver bullet to all change problems. But as it turned out in the particular case of UNO, the lack of double loop learning was one of the biggest constraints in UNO´s Kanban system after a few months. Professionally facilitated, regular retrospectives helped to overcome this constraint. In concert with a deeper understanding of evolutionary change through Kanban classes for all UNO team members, these retrospectives encouraged the self-confidence to jointly master the current challenges. Additionally ongoing leadership training and coaching helped managers to change to be more service-orientated and support for their teams. Leadership was more and more separated from the hierarchical order and became a kind of team sport that considered everybody to be an essential part of the game (see “Leadership as a team sport”)
All these insights and improvements could have happened earlier. Likewise, the risk of gradually fizzling out could have been reduced. The fact that the leadership team was able to face its own shortcomings, nevertheless, contributed essentially to the current state of UNO´s Kaizen culture. Eventually, it is at the heart of this culture to not just allow failure but to make it a catalyst for learning and improvement.
About the Authors
Michael Beyer has worked in IT for over 15 years, in various roles as Software Developer, Architect and Project Manager. Since early 2012 he is Department Manager for Unified Network Objects at Swiss Railways (SBB AG), Berne.
Siegfried Kaltenecker is managing director of Loop Consultancy, Vienna, specializing in Systemic, Lean & Agile change management. Sigi is editor of the “Platform for Agile Management“ (p-a-m.org) and co-author of “Kanban in IT. Creating a culture of continuous improvement”.
Thanks to Ben Linders, Klaus Leopold, Marius de Beer, Cliff Hazell and Peter Hundermark for their valuable feedback.
- Anderson, David: Kanban. Evolutionary Change Management for Your Technology Business. 2010.
- Leopold, Klaus; Kaltenecker, Siegfried: Kanban in IT. Creating a culture of continuous improvement. To be published in English in May 2014. If you are interested in a draft version please contact firstname.lastname@example.org or office@LEANability.com (German version: Kanban in der IT. Eine Kultur kontinuierlicher Verbesserung schaffen. 2. Auflage. München 2013)
- Platform for Agile Management
- Reinertsen, Don: Flow. The Principles of Product Development Flow: Second Generation Lean Product Development, Boston 2009.
- Research groups on most important insights
You are comparing apples and oranges. Scrum doesn't have a project manager role and PMP looks like the route for the traditional project manager.
I can see PMP being useful if you are working for traditional companies. On the other hand, I think Scrum training is all about the certification treadmill and the associated fees - a 2-day Scrum Master course is a joke.
Just so you know we've removed the PMP comment you had replied to and have also blocked this user. She's been a fairly persistent spammer advertising Scrum training courses!
Head of editorial, InfoQ.com
We want more real life case studies
I think that real life case studies are a great way to learn what works and what doesn't. They can inspire people and give them ideas that they can use in their daily work.
If anybody has a case study that they would like to share on InfoQ they can contact us.
Editor on Agile at InfoQ.com
Srini Penchikala Aug 21, 2014