Recap
In the first post we started to follow the journey of a project from chaos to control. We value some kind of structure and certainty, and right now things appear to be largely unmanageable. The people on the project are struggling with a creative piece of work that seems to be in perpetual flux, with unclear requirements and unpredictable outcomes. In addition they have "firefighting" of one sort or another to deal with, and team members feel as though they are being reactive rather than proactive. They are overwhelmed by too many decisions, and too little information or time to make those decisions well. Any planning meetings they try to have, for example, are made unproductive due to constant crisis management. Priorities are unclear and the initiative has stalled. The team senses that they are powerless, unable to plan or forecast, or to offer reassurance to stakeholders. They are immersed in a shifting situation that is devoid of any clear pattern, and with little opportunity to develop teamwork they hardly feel like a team at all. They seem to be a million miles from achieving control.
Half an hour towards hope
If we are to gain control, then we have to be able to relate cause and effect. Bear in mind that the three legs of empiricism are transparency, inspection and adaptation. Transparency is bound to the Scrum value of openness. We're after a consensus view of reality, without which inspection and adaptation can never really occur. In other words, people must first be in agreement about the things they actually see happen.
A facilitated half hour workshop can help build up a head of steam. Start with an individual writing exercise, for example, in which each team member describes -- on index cards or similar -- the things he or she has been working on today. A time-box of no more than ten minutes ought to be allowed for this part of the exercise, and only one thing should be written down per card. For each task so identified, encourage the author to then further break it down into constituent steps or sub-task, some perhaps already completed and others perhaps not yet started...again, one per card. Continue until a granularity of no more than a day's effort to complete each item is achieved, or until the ten minute time-box ends. Expect no agreement between participants at this stage on how work is either being identified or structured.
Next, group everyone's items together and ask participants to affinity map them. Which items are actually the same, and have multiple people working on them? Whose card expresses the activity best? Can any duplicates be quickly rewritten into something better? Allow another ten minutes for this.
Now to start putting all of this in order. Arrange the cards under three headings: “To Do,” “Doing,” and “Done.” Ten minutes again for this activity. Although it may look like a Scrum or kanban board, the resemblance is only superficial. As of yet we have neither explicit policies in place nor an economic model of production. For the moment, we are just establishing transparency -- that consensus view of reality -- over the current state of play. From this, policies and models may emerge later, if we allow them to, and learn to listen.
The layout
To do: This column lists those things yet to be worked on and remain unstarted. You can think of it as a backlog of tasks, and a rough attempt can be made at ordering items by priority.
Doing: This enumerates those tasks actively being worked on right now. Count them up to gain an understanding of the current Work in Progress (WIP). We will eventually limit this number to improve focus and reduce the waste caused by trying to juggle too many things at once.
Done: Any tasks thought to have already been finished. This provides a visible record of progress, and team members can find it motivating to see evidence of the things they have achieved.
You can use paper and sticky notes for laying these things out, a whiteboard, or equivalent online tools. Remember to keep the board updated! Keep on refreshing the To Do column (or "station") with new tasks as they emerge, roughly sorted by importance or urgency with the ones to do soonest at the top. Now let's think about limiting the amount of items in progress so they can be finished and moved to Done more quickly. We need to stop starting and start finishing!
Limiting WIP
The first attempt to limit work in progress is likely to prove difficult and uncomfortable. In most organizations there is a cultural pressure to start work and to be seen to be busy. It almost feels like there's an obligation to put every task under the “Doing” column, just as a political signal to show that it has been neither ignored nor forgotten. After all, it's more likely that team members will be criticized for not starting a piece of work, than it is that they would be commended for finishing something well. In truth of course, there is a limit to how much people actually can juggle at any given moment. The immediate challenge lies in baking transparency over actual work in progress into the system.
The "avatar" technique might be used, for example. You might give each team member exactly one "avatar" each to represent themselves, and which they can then attach to the item they are genuinely working on right now.
Ask each team member: What actually are you working on right now? What have you put to one side and will have to return to later? It will become clear that no-one is truly handling more than one task at a time, regardless of how much they have started. Other things can held in the “To Do” column, and rewritten if necessary to better express the work that genuinely remains.
- A system like this makes work visible. Team members can see at a glance what work is next in line, what they’re in the middle of, and what they’ve completed.
- People are less likely to be overloaded. Establishing work in progress limits help conquer the habit of starting too much, and finishing too little.
- The quality of the work is enhanced because of improved focus and attention. Rework is reduced.
- Less time is wasted switching context.
- Motivation is given a boost. Looking at work in the Done station reminds team members of what they've achieved.
- Baby steps! Although the technique is simple and seemingly trivial, it can be difficult to break old habits and to make this work.
The conversation around limiting WIP involves making “now” crystal clear, protected, and rewarding. Here's how a foundation of personal discpline might be built before work is even put on the board.
- Choose one current task and name it. Use a short sentence, e.g. “Right now I am refactoring the customer activity log handler”
Put that task immediately in line of sight, on a sticky note perhaps - Remove competing options from view (actually hide the long to‑do list, close open windows and tabs) so there are fewer distractions.
Use timeboxing:
- Begin with a 15–25 minute timebox in which a team member commits to working on one task until the timebox expires. Then a short break of 2-3 minutes might be taken.
- The objective is to achieve focus until the timebox ends...not necessarily to finish the task in that timebox. This lowers resistance and helps to build flow.
- Write the next task taking into account work that remains. Scope the task realistically, based on the evidence of the last timebox, so it is more likely to actually be achieved in the next one.
- After disciplined focus is established, the item can be put on the board in the "Doing" column
The situation so far
At this point we have a To Do list. That's all. It may not seem like much of an achievement, but using a To Do list properly does require discipline. That's where we've won: an honesty and transparency about work in progress is beginning to crystallize. Most organizations struggle with the basics in their agile journey, and then attempt to scale precociously. Yet without this foundation of good agile hygiene and practice, it ends up being their existing dysfunctions that are "scaled".
So, we can see the beginnings of a managed workflow starting to take shape. Predictability and flow just about seem to be within our grasp, and perhaps rough forecasts might be made about delivery dates.
- Having clear policies around To Do, Doing and Done improves forecasting. It turns cross-your-fingers-and-whistle guesses into data. We can reasonably say, for example, that work items of a certain kind might take 2 or 3 days.
- Each time a card progresses into Doing and then Done, we get a real data point. Over time, we can use aggregated data for evidence based forecasting.
- Focus is promoted. By working on fewer things at once, progress becomes steadier and more predictable, and cycle time can begin to be appreciated.
- The likely impact of an increased workload on productivity can be determined.
- A disciplined "Doing" station encourages openness about how busy the system is right now. There is an opportunity to frame commitments which are more realistic.
In the next post, we'll tighten up our understanding of what it really means for work to be Done.