Are you always short on time and budget, even though your team is motivated and skilled? Do you struggle to keep up with ever-changing demands from your clients? Do you often find yourself being unable to deliver on your commitments? We know these feelings because we’ve been there. This is the story of how our team at Nave turned our Kanban process around and increased our productivity by 250% in less than two months.
Managing emergencies
In a fast-paced environment, clients want results yesterday – it’s easy for every request to feel like an emergency. Too many emergencies mean lower priority tasks are neglected, decreasing overall productivity. We struggled with this effect, always striving to go the extra mile to exceed client expectations. We introduced Classes of Service in our process – rule-breaking emergency tasks were limited to one at a time.
Breaking the cycle of flow debt
We knew we had bottlenecks within our process – but what was the cause? Our first step was to examine our Aging chart. The Aging chart shows the age of tasks that are still in progress, whereas the Cycle time scatterplot shows the cycle time of completed tasks. Tasks in progress that are older than the cycle time on the 70th percentile are likely to be holding up your process – the older the tasks, the bigger the chance of delay.
From looking at our Cycle time scatterplot, we knew plenty of tasks were getting delivered in a predictable manner. The Aging chart, however, showed us that our fast cycle times were lying to us. We had a clear bottleneck in the Testing state – we didn’t have the capacity to handle the upstream workload. Our QA guys were always picking up small, fast tasks in order to release work faster. Some of the older tasks that were too heavy or complex aged even further. This approach was generating large amounts of flow debt.
We had to complete the oldest tasks to get back on track. This generated even more frustration. Developers had a lot of re-work due to merge conflicts. QAs were constantly reporting severe regression effects. Once the old tasks got completed, the problem revealed itself right away in our Cycle time scatterplot – our average cycle time increased dramatically. The management thought the project had span out of control. But it hadn’t. That was the cost of the huge flow debt. We paid it off, learned the hard lesson, and decided we’d never let this happen again.
Setting explicit pull policies
In order to stop flow dept from building up again, we had to introduce pull policies for every process state. It was essential that our team works on tasks on a first-come, first-served basis. Every task that goes to the next process state is placed at the bottom of the column. Once the team has the capacity to handle new work, they pull in the task at the top of the previous column. Tasks were no longer delayed or abandoned because each work item was processed in the order it arrived – regardless of its difficulty, size or priority.
This does not mean that we stopped prioritizing our tasks. Prioritizing is very important when trying to decide which task should be the next one to be pulled in to the process. However, prioritization takes place and Classes of Service are assigned when items are added to the To-Do list. Once the team has pulled a task into the process and committed to the item, the priority is no longer taken into account, and tasks are dealt with in the order that they arrived in the process state.
With Kanban, you can see the results of your changes immediately. We first saw the difference in our cycle time scatterplot – our cycle time was 250% faster than the yearly average cycle time. After some time we started to see the long term impact of our changes.
Our Cumulative flow diagram showed smooth, inclining bands rather than flat areas and sharp jumps. With flow debt eliminated, our process became consistent and predictable. This helped us increase the accuracy of our estimates. Lastly, as a team, we felt happier, more motivated, and in control of our process.
The Nave’s versatile analytics suite makes the bottlenecks and problems holding up your process easier to find and eliminate. You don’t need big changes to see big results – something as small as changing a pull policy can increase your productivity by 250% in less than 2 months. The trick is to pay attention to your efficiency and apply small continuous improvements.
Do you use Kanban in your process? How do you assess your productivity? What problem areas did you discover? Tell us about your experience in the comments!
