Facilitating the Spread of Knowledge and Innovation in Professional Software Development

Write for InfoQ


Choose your language

InfoQ Homepage Articles Using the Kanban Canvas for Driving Change

Using the Kanban Canvas for Driving Change

According to Karl Scotland the need for learning organizations is greater than ever. People need to be able to continuously solve new problems, they have to develop thinking and problem solving skills that would enable them to do this.

In the interview with Karl Scotland at Agile 2014 he introduced the “kanban canvas”: a problem solving model using kanban-based techniques. Karl did a workshop at the Lean Kanban France 2014 conference where he explained the kanban canvas and showed how it can be used to drive change in organizations.

The kanban canvas helps to have to look at things from a different perspective. It uses storytelling to understand the current process and create a collaborative insight how work is flowing. People can use it to study the system to learn more about the current situation, describe desirable impacts and define interventions to realize those impacts.

In an interview with InfoQ Karl explains the kanban canvas and explores how it can be used to create shared insights and decide upon the approach to intervene in organizations.

InfoQ: Can you describe how the kanban canvas looks?

Karl: The canvas is a single page, designed to be printed A0 size (33 x 47 inches) so that it can be used as a focal point for collaborative work. The page is divided into several sections which focus on different aspects of designing and running a kanban system.

(Click on the image to enlarge it)

In the centre is an octagon, with eight associated segments radiating around it. The central octagon is where an understanding of the purpose of the system is built. What is the problem are we trying to solve with a kanban system?

On the right, three segments flow out of the central system, representing three perspectives to look at the desired impact we want the kanban system to have; flow (doing the thing right), value (doing the right thing) and potential (doing the thing sustainably).

On the right, five segments flow into the central system, representing five approaches to making system interventions which we hope will have a positive impact; studying the context, sharing the understanding, stabilising the work, sensing the capability and searching the landscape.

InfoQ: The central part of canvas is a systematic view of the problem or situation that is being addressed. What kind of techniques can be used to described it?

Karl: Systems Thinking uses the iceberg metaphor, where individual events are what we see above the water line. Below the water, often unobserved, are the patterns of events over time. Below that are the system structures that create those patterns, and even further down are the mental models that lead to those structures. To begin to understand the systemic problems we want to address, we first need to look for the patterns.

One way of finding patterns is through narrative. I ask people to tell stories about what has happened in the past, over time. Who are the key characters in the story? What was their situation? What key turning points occurred? What were the consequences? In summary, what has happened in the past that has led to the current situation.

A quick and simple technique to do this is using a format known as the Pixar Pitch, a formula which can be used to summarise any Pixar film.  It looks like: “Once upon a time … Every day … One day … Because of that … Because of that … Until finally ...”. Collaboratively filling in those blanks can create a common understanding of the challenges being addressed. I like to say that the final “Because of that” is “we are designing a kanban system”, and the “Until finally” will be described by the impacts we want to have in terms of flow, value and potential.

InfoQ: In the article Kanban - isn’t it just common sense? you explained the role of heuristics in kanban thinking. Can you explain why you prefer heuristics over rules for problem solving?

Karl: Rules are meant to be followed, and people often blindly follow them without thinking. Either that or people find ways to break the rules without getting caught. Heuristics, on the other hand, are rules of thumb which lead to discovery and learning. So by focussing on heuristics, my goal is to develop problem solving capability, rather than define any specific solution.

This is particularly relevant given the complexity of the world nowadays. Rules are a good approach when solutions are proven and repeatable, but this is rarely the case when we have new teams, developing new products, in new categories and markets, with new technologies.

For organisations to be resilient to rapidly changing conditions, they need to be learning organisations, able to respond and adapt quickly. They need to be able to have eureka moments. I like the fact that the word heuristic comes from the same root as ‘eureka’, and we seldom have eureka moments about something we already knew.  These moments come from abductive reasoning, or the logic of ‘what might be’, rather than inductive (‘what is’) or deductive (‘what must be’) reasoning which rely on past data and evidence.

InfoQ: The canvas explores the impacts, looking at flow, value and potential. What makes these things so important?

Karl: They evolved that way in response to some of the conversations I have had about using kanban systems, and people’s perceptions of them. Many thought that the focus was purely on optimising the process, or the flow of work. However, we also know that flow is not sufficient if the work product has no value. And also, a flow of value will not be sustainable if we don’t consider the people doing the work - the system will have no future potential,

It turns out that those three perspectives - flow, value and potential - form a triad which helps encourage a balance between the three. All three are necessary, and none of them are sufficient on their own. The triad helps people think about the affinity of a potential future to each type of impact, without having to make a choice, or become overly focussed on one.

InfoQ: You talked about why fitness for purpose is important for the changes that we want to do. Can you elaborate?

Karl: Kanban systems provide an evolutionary approach to change. That means that rather than defining a single future state, and working towards that, we start by understanding the current state, and exploring the many potential future states from where we are now. However, that means that while we are exploring, we need a way of assessing whether a potential future is one that should be pursued further, or reverted from.

Fitness for purpose is a way of making that assessment. Identifying the purpose of a system, or an organisation, and from that deriving appropriate measures of fitness, enables a range of outcomes to be explored and evaluated. Similarly, by continuing use the narrative approach, we can define fitness for purpose as having more stories related to positive impact, and fewer stories related to negative impact.

One way of thinking about it is that fitness for purpose sets the general direction, without being specific about the destination or route.

InfoQ: To drive change in organizations interventions can be used. Can you explain how the kanban canvas supports them?

Karl: The canvas has five ways of looking at system interventions; study, share, stabilise, sense and search. These 5S are abstractions of the various practices and techniques we can use to act on and influence the behaviour of the system. The names are chosen to describe the intent behind what we do (as well as because they all begin with the letter ‘S’!).

  • Studying the system covers practices which help get necessary knowledge about the context before we start making changes and evolving.
  • Sharing the understanding covers approaches to visualising the work and radiating information to create a common appreciation of the situation.
  • Stabilising the work covers ways of creating explicit limits and policies which form enabling constraints, become a baseline for improvement, and build resilience and adaptability.
  • Sensing the capability covers both quantitative and qualitative ways of getting feedback on how well the system is performing, through a mix of metrics and meeting cadences.
  • Searching the capability covers the running of experiments and other approaches to generating learning, evolving and improving the system.

InfoQ: You mentioned that studying the context can help us to understand how to do effective change. Can you give some examples?

Karl: There are three main areas that we can study to learn more about the context we are wanting to change; the customer (or stakeholder), their demand and the workflow we use to satisfy that demand.

Learning more about customers and stakeholders means having empathy with them by spending time with them, experiencing their frustrations and fulfilments. Empathy Mapping, to capture what people are saying, doing, thinking and feeling is a useful technique for this.

Learning more about the demand means looking at the nature of the work that is requested. Stephen Parry’s CORE profile is one way of looking at the value of work. What Creates value now? What provides an Opportunity to create value in the future? What is Remedial work, fixing value already created? What is External work, fixing problems created elsewhere?

Similarly, Cost of Delay and Classes of Service provide a way of looking at scheduling and responding to work. What are the revenue or cost implications, and what is the urgency?

Learning more about workflow means looking at the way the work is done. Value Stream Mapping, or equivalent variations for knowledge work, identify key process points. Where are there delays, usually in the form of queues or batches? Where do we need knowledge to make decisions, such as what to build, how to build it, or whether it is done? Where are there feedback loops that are negative, causing delays, or positive, creating feedback?

InfoQ: The canvas uses experiments as an approach to change. How does this work?

Karl: The evolutionary approach that kanban systems take means that we can’t be sure exactly what the future system and process will look like. The canvas’ experimental approach, therefore, is a way of searching around for potential changes and testing them.

By using the various perspectives given by working through the canvas, teams can come up with hypothesis about what they could do differently, try it out, and either validate or falsify their idea. The metrics and feedback cadences discussed for sensing the capability become relevant here.

Another important consideration is making the experiments safe to learn. The nature of evolutionary change means that not every experiment will give the anticipated results, so any changes should be easily reversed or otherwise dampened in that case.

In some ways this is similar to the Lean Startup idea of running lots of small experiments, except with the kanban canvas we are running experiments more to do with the process than the product.

InfoQ: Can you give some examples of experiments that you have done in organization? Which ones were successful? What did you do when the experiment failed?

Karl: When I first started experimenting with kanban systems in 2007 at Yahoo!, the early changes were mostly around different cadences. At that point the experiments were less structured and disciplined so the results were more subjective. We moved from a typical 2 week timebox, or what I would call a single metronomic cadence, to a more poly-rhythmic cadence, with the various meetings decoupled from each other. We would schedule work at the start of every week, plan it on demand when we were ready to actually start the work, review progress at the end of every other week, release in the middle of every other week, and retrospect every month.

More recently, I have worked with an organisation that wanted to be more reliable in predicting delivery dates. We used an A3 format to work through the problem, background, hypothesis, details, safety, measures and plan. The hypothesis was that by capturing and analysing Lead Time and Throughput data, they would be able to achieve a desirable Due Date Performance level. A certain percentage of work would be delivered by the predicted date. The safety was achieved by keeping the data internal to themselves before using it to make any external commitments.

I tend to think about experiments in terms of learning rather than failure. The only failed experiment is one which isn’t run to completion! A consequence of that is that I find that learning leads to continuing to moving forward rather than back. Experiments tend to get tweaked based on the learning and feedback as part of the evolutionary process.

InfoQ: What kind of facilitation techniques can be used when applying the kanban canvas?

Karl: One of the core techniques that is used with the canvas is that of creating divergence, before convergence, and resisting the natural temptation we have to jump into solutions. The reason for starting with understanding the systemic problems, and desirable (or undesirable) impacts is to deliberately create this tension.

I also like to work with groups by taking them through an exercise multiple times, beginning by thinking individually, and then combining ideas into pairs and groups until the whole team has an agreed answer. This is a technique I learned from Jean Tabaka at Rally as a way of giving everyone the opportunity to have a say, and encouraging a diversity of ideas while still working towards team consensus.

A similar approach, particularly when designing experiments, is Cognitive Edge’s Ritual Dissent. With this technique, critical listening and feedback to a range of ideas leads to more coherent results.

InfoQ: If people want to learn more about the kanban canvas, what can they do?

Karl: The main place to start is the canvas website which is where the pdf can be downloaded from, and where I am building out content around it. There is also a lot of related writing I have done on my blog.

I have also been offering to have a short call with anyone who downloads the canvas to help answer any questions and explore how it can be used. Just say you want more information when you submit a download request and I’ll get in touch!

Finally I do have a workshop on Kanban Thinking which I can run publicly, or privately which spends 2 days working through the canvas, digging into many of the areas in more detail.

About the Interviewee

Karl Scotland helps businesses become Learning Organizations. Over the last 15 years he has been an advocate of Lean and Agile approaches, working with companies including the BBC, Yahoo!, EMC Consulting and Rally Software. He has been a pioneer of using Kanban Systems for software development, a founding member of both the Lean Systems Society and Limited WIP Society.

Rate this Article