NOTE: I’VE REVISTED THIS TOPIC MUCH MORE RECENTLY, IF YOU’RE INTERESTED ON HOW MY VIEWS HAVE CHANGED READ 4 TYPES OF WORK REVISTIED.
If you have read The Phoenix Project (which I highly recommend you do) you’ll be more than familiar with Eric who introduces us to the four types of work.
- Planned Work – these are typically business projects or new features
- Internal Projects – server migrations, software updates and so on
- Changes – usually driven by feedback on already completed work
- Unplanned Work – support escalations and emergency outages
In the book Eric talks about how Unplanned Work is unpredictable and destructive as it can often push out planned work. I can’t even begin to count how many times have I’ve planned in projects and tasks only for a crisis to arrive and my sprint to fail!
Managing a Product Backlog is an excellent way to handle three of the four types of work. After all, Agile Principles encourage us to collaborate with stakeholders, listen to feedback, and embrace change. The value of internal projects can be evaluated against other projects and then planned correspondingly and what is the backlog at all if not a prioritised list of Planned Work?
It’s the infamous Unplanned Work which continues to disrupt our Sprints, break our commitments and lower our productivity. Over a few years I’ve tried a number of different approaches to handling the inevitable Unplanned Work.
I’ve tried introducing slack into everyone’s sprint, only planning fifteen instead of twenty Story Points. I’ve tried allocating Support Tickets as they come in, assuming that the demand will cause our average sprint velocity to drop and the balance between Planned and Unplanned Work to find a happy equilibrium. It never does!
The only strategy I’ve found to work is to completely isolate the Unplanned Work from the other three types. Create a specific team (in our company we call this the SWAT team) who will react to all incoming unplanned work and “pitch in” with other work when there isn’t enough to do (Parkinson’s Law pretty much guarantee’s that this never happens).
The SWAT team handle the urgent, unplanned, and reactive work (often using a strategy like Kanban or Scrumban) while the rest of your team focus on delivering maximum value to the business (Sprints, commitments, and all the other wonderful concepts we hear about on Agile courses). That’s not to say they don’t fix bugs or take on internal projects, it’s that they work in a structured, planned environment in an effort to build solid releases and reduce the amount of Unplanned Work long term.
Speaking of Parkinson’s Law if you do decide to create this buffer between Planned and Unplanned Work (and I strongly suggest you do) you need to consider what happens when the work expands to fill the allocated time. What happens when the urgent support demands of the business exceed the capacity of your SWAT team? Hire extra people? That would be nice! But I can tell you what you must not do – DO NOT pass the extra work over into your Scrum team. I’ve made that mistake before and I can assure you that it’s a very slippery slope which is nearly impossible to come back from. Once you start roping in extra developers you undo your good work and Sprints start to fail once again. I mention it between I’ve seen it happen!
I hope I’ve given you some thoughts, I hope that my views on throttling the amount of damage your Unplanned Work can do help you safe your own commitments but I’d be delighted to hear your views – what do you to do safeguard your team’s productivity while meeting the urgent needs of the business? Post a comment below and let me know!
4 thoughts on “The Four Types of Work in an Agile World”
Interesting blog, the SWAT team sounds like a good approach. I work in an environment with over 30 domain teams and the SWAT approach failed for us, this was largely due to the team working across the whole system and often in areas where the functionality and code base is something they are unfamiliar with.
We implement a time box approach but I feel this has its limitations. The approach is to reduce the teams velocity by 10%, is it then calculated as an amount of time (depending on velocity and available team members). My approach would be to keep this concept as we include it in projections, then the time would rollover where it is not used up until the point where you have planned out your future sprints. Ideally plan three sprints ahead, anything beyond that leave as an Epic as detailed analysis too far in the future will take time and has a significant risk of change (if we are doing scrum correctly).
You cannot plan for everything and sometimes you need to perform an abnormal sprint termination. A key here is to ensure your stakesholders understand this and accept the possibility, this way it won’t be a surprise. If nothing is a surprise then it will be better received even if the news is not great.