The cost of context switching
I joined a new journey a few weeks ago. I was asked to improve things slowly as the teams were performing pretty well for the last months. I accepted the challenge as I saw a lot of things were implemented well and I just changed one thing. I asked the teams members of the different teams to be aware of context switching. The result is an increase of 20% of the velocity in just a few sprints.
Individual Level
Scientists have demonstrated with fMRIs that individuals cannot perform more than one task at the same time. Only impulses in each different hemisphere of the brain may happen at the same time. Studies demonstrate that IQ test results vary with and without interferences (phone calls, emails, etc) for the same people. The results decrease in 10 points when people are interrupted while performing the test in comparison to performing the test without interruptions.
On the other hand, if we try to analyse how computers multitask, they execute the tasks one after the other but very fast. So fast that we have the perception things happen in parallel. For example, different applications running at the same time, etc. The true is computers can do small operations for each of the applications we have opened very fast but humans cannot do the same. If we try to switch from one small task to another, there is a cost intrinsic in moving between them. For medium or complicate tasks this cost is even higher. As a consequence, we cannot do the same than computers. We need to focus on tasks for some time. Especially when we are performing complicate tasks, we need some time to really get the understanding and context to get the job done properly.
Team Level
Collaboration between team members is fundamental. We all know that. It is one of the foundations of Agile. So, how can we balance the reduction of context switching with the necessary collaboration between team members? This is something that may vary between teams and companies. Something which is working in this journey is the use of messaging apps like Slack, HipChap, Campfire or any of the others. These tools allow team members to ask for anything they need from their peers who may have a look to requests on real-time or later depending on their load at the moment. We encourage team members to unblock their peers as soon as possible as explained in this article about the engineering culture in Airbnb.
Epic Level
Most of the companies love to work in several epics at the same time. Everyone believes that of all the epics are important so we need to demonstrate progress in all of them at the same time. The reality of this is that, as demonstrated by thousands of the epics studied by Yale University, the epics end up delivered almost twice slower than if they were prioritised and serialised reducing context switching for the teams. More about this in our article Priorities based on the economics view of flow.
We found really interesting the experiment by Harvard Business Review that prove the statements above at The Multitasking Paradox.
So, what is your experience with that? Have you been in a similar Agile Journey? What were the main takeaways from your journey? Please share in the comments below… 😃
