Storytelling with Scatter Plots
Are Scatter Plots Able to Tell a Story?
Scatterplots are a visualisation to show the relationship between two numerical variables. In Flomatika we display lead time over time. They can be used to conduct deeper analysis, but today I want to show some examples on reading scatter plots just by recognising patterns that emerge from them. How these patterns are able to tell a story.
A few things to note before we dive in. Although these scatter plots are ‘inspired’ by real charts we have witnessed, they have been stylised for the purposes of this blog. No dots were harmed in the making of this blog.
A new Initiative has been spun up with a newly formed team. Greenfield development, a clean slate of a backlog to shape our own destiny and no legacy system to maintain. Initially, things look great. Plenty of items are getting done with relatively low lead time.
However, what seems to be a promising start, has slowly spiralled out of control. We didn’t keep an eye on our WIP, and were only looking at the completed work. The team kept starting more than they were finishing. Slowly but surely, the pieces of work that were in WIP for a long time start to be completed, and they show up on the scatterplot.
Each of these more complex pieces of work takes longer and longer each time to be completed. This is compounded by some bad practices adopted by the team. As things take longer, requests are ‘expedited’ to try and beat the long lead times. This results in the existing work in process being placed on pause. A start and stop rhythm is establish. Each restart taking a longer time to ramp up. Now all the strategic pieces of work take such a long time to complete. The cycle repeats each time only getting worst. Entering a death spiral.
Big Batches vs Continuous Delivery
Comparing two scatter plots can also tell a story. The first is a group that releases in big batches. Although the individual work items have been developed by the team, their value is not realised until they have made it into production into the hands of their customers. Consequently you’ll see these vertical clusters of dots whenever a release is done. Typically at a regular cadence, with some vertical clusters denser than others.
Compared to this second group, who practises continuous delivery. Items are released into production into the hands of their customers the moment they are completed. A pattern we’ve seen is almost like '5-fingers' for the week. The teams are not releasing on the weekend, so there is a gap between the weeks. Also a typically pattern across teams is there are more things done mid week, compared to the start or end of the week.
The Big Clean Up
A pattern we see very often with our new customers is the ‘big data hygiene clean up event’ that occurs. When teams start seeing their data visualised, it’s also easier to see where the bad data hygiene is. Which in turn makes it much easier to find those items and do a bit of a spring clean. So we often see this big ‘pillar’ as these old forgotten items are categorised. It also acts as a nice “line” to show life pre-visualisation and post-visualisation. What was the lead time like before, and how the teams have improved since.
Then there are some charts that scream out that a major event has occurred, and it completely changes how the teams are operating, and it becomes clearly reflected in the scatter plot.
In the chart below, this team was clearly doing big batch releases, but their lead time was also steadily increasing over time. A team reset occurred (nothing was completed for a short period), then the team restarted. They have less ‘batching’ of their releases, and their overall lead time has dropped, and continues to come down. They are still some long lead time items occasionally, but these are probably legacy work from before the team reset point.
You can practically draw lines on the scatterplot to annotate what is happening. In this example we can see a steady increase of lead time (red line); before a reset was taken (yellow area); and when the team comes out from the reset, their overall lead time has a downwards trend (green dotted line).
This last one might be a bit harder to spot. What you see is a team that has gone through three phases of disruption, where the lead time has jumped up due to some external factor that they needed to respond to. They tackled each of these the best they can under the circumstances, but there is no denying that lead time has been visibility impacted.
However, it should be noted that for each disruption, the 'peak' lead times have been lower. What this means is that the team have shown to be more resilient with each disruption they faced. The cause is likely outside their control, but each time they have been able to face it, and reduce the impact each time.
If we add some lines to the scatter plot, you can see the patterns easier. The results of the three 'disruption' events are represented with the yellow dotted lines. Each subsequent disruption, the 'peak' of their Lead Time has been lower each time. Between these 'disruption' events, you can see the teams are still overall reducing their lead time over time. A sign that the team is not only getting better, but are also more resilient to disruptions.
Scatter Plots Are Able to Tell a Story
Not all scatter plots will be able to tell you a dramatic story. Occasionally you’ll see one that makes you pause and appreciate what it’s able to convey. They are also not the easiest thing to read, and may require some experience to extract the story out of it.
Sometimes they answer some questions: The team’s focus on reducing their lead time is bearing fruit, and we can see them trending in the right direction.
Other times, they allow you to ask some questions: Something happened at this time and it has disrupted the team. What was the event that happened at this time?
If you haven’t before, hopefully now you can appreciate how scatter plots are able to tell a story. What does yours tell you
These other blogs might interest you:
- Service Delivery Review: Speed
- Make better day-to-day decision by having visibility of your WIP Age % vs SLE
- Lead Time Predictability: Is your team predictable?