Everyone knows that keeping Jira data tagged consistently is difficult at all levels of an organization. Some teams are on company-managed boards with required custom fields, while other teams use team-managed boards where those same rules don't apply. Some teams are on the cloud, and others running on-prem.
One consistent problem we've heard from engineering leaders is that they can't create a unified view of their data that lets them understand their investments without costly initiatives to clean up Jira.
For this reason, Uplevel created the Machine Learning (ML) Issue Classifier.
To find your team's top-level investments (New Value Creation, Sustenance, and Defect Work), head to the Plan page and select the "ML Issue Classification" rule.
From here, track how your team's time investments to answer questions like:
- Are we investing enough in new feature work? Track how New Value Creation has trended over time and set a target for your organization.
- Why is New Value Creation low? Is it because too much work is going towards Sustenance like Keep the Lights On projects? These tend to quietly detract from an organization's focus because the team gets "used to it."
- Has defect work started to creep up over the past quarter? This may be a sign the team in need of some dedicated tech debt reduction?
💡 Example:
For this team, the past month been a push to get several new features out the door and into customers' hands. It's great to see that New Value Creation is the top level investment with 32% of time going towards that. However, we want to keep an eye on work on Defects (19%) and Sustenance (11%), which are comparable to New Value Creation when combined.
This period was planned to be mostly new feature work for this team, so it's a bit concerning to see that Sustenance is taking up 11% of the team's time. To see where that investment went, we then click into the Sustenance category. We see that it's on-call work that's comprising the majority of the Sustenance work, followed by an ongoing infrastructure project that should be wrapping up soon. It's a surprise that that on-call was taking up that much effort, and merits a conversation to see if there are opportunities to reduce the on-call load and automate some of those processes.
Learn more about how ML Issue Classification works: