How do you start with a product backlog when you’re transitioning to agile? In this article, Darin Kalashian shows us how a cross-functional team at the product owner level creates a product backlog.
When a product development team begins to use the Scrum framework, one of the first objectives is to form a product backlog for the Scrum team to work off. The product backlog is the product owner’s responsibility to oversee, though the product owner seldom has all the knowledge needed to define a high-value solution that will be successful in the marketplace. To create a useful backlog, the product owner must rely on supporting roles, such as system architects, management, and subject matter experts.
The Scrum Guide by Ken Schwaber and Jeff Sutherland, the originators of Scrum, lightly touches on how to groom the product backlog. As an experiment, our development team complemented The Scrum Guide with some additional events and artifacts to help ensure product grooming was performed to a level where sprint teams could go to the next step, sprint planning. Additionally, by inserting a definition of done for the grooming of product backlog items, we made it easier for the team to manage the quality. This article defines the high-level workflow used to streamline our execution of product backlog grooming.
Where to Start?
Just like every other team out there that has had to get ready for its first Scrum project kickoff, we did all the preparation—reading through The Scrum Guide, enrolling in formal Scrum training to better understand the fundamentals, and assigning the roles of product owner and ScrumMaster. We huddled back in the office, all literature’d up. We even organized a Sprint 0 to get everything set up and ready for kickoff. We thought we had a handle on it ... until we sat down with our product owner, sprint team, ScrumMaster, and various stakeholders and tried to make a plan out of our product backlog. We quickly realized that grooming our identified product backlog items wasn’t as easy as we first thought.
In our environment the application is very complex, with multiple vendors, subsystems, and customer needs, not to mention our own roadmap and business objectives. It was unrealistic for the product owner to know it all. For the product owner to understand the business value of a customer request, we needed to pull in a larger, cross-functional team of system architects, business leaders, and others not typically part of a “traditional” grooming session. However, we agreed that having a grooming session with all of these roles present in addition to our development team would not be particularly productive.
As the coach of this Scrum team, I suggested that all of our stakeholders should huddle up and brainstorm some ideas on how to make product backlog grooming more effective. This collective team decided to experiment with breaking the product grooming activity into two separate, supporting activities. We termed these activities “product backlog planning” and “product backlog grooming session,” as depicted in figure 1. While we did not want to squelch the team’s emphasis on collaboration, we needed to make sure each contributor maximized his or her value and use of time.
Figure 1. Product Grooming
Product Backlog Planning
Everyone agreed the product backlog was not ready to be groomed by the Scrum team. There were too many unknowns. The Scrum team wanted backlog items they could process and proceed with for implementation. As it stood, the product backlog identified many items that, as the measuring stick would tell, were of epic size. The product owner was given the task to meet with knowledgeable system architects, business leaders, and technical managers in an attempt to flush out the customer needs and capture as many story points as possible to convey the functionality. This became the goal of the product backlog planning sessions. In these sessions, the product owner—along with the system architects and leaders—defined feasibility, required high-level functionality, and product roadmaps that would support implementation. In the spirit of Scrum, a definition of done was established to identify when a product backlog item had completed planning and was ready for the development team to size and consider for a sprint.
The definition of done identified these acceptance criteria:
- Planned backlog items: Backlog items needed to be broken down into stories that were understandable to the development team. Epics were clearly identified and tracked. But as always, epics weren’t ready for grooming until there were supporting story points.
- Rough order magnitude of size: While the development team would provide an estimate of the story point later, the planning team was asked to identify a T-shirt size estimate for each backlog item. While it had no direct value for story point estimation, it did help justify the business value logic.
- Business value: We found it useful to carry a business value index separate from the backlog item priority rank index. The business value was the justification, or backup data, to support the backlog rank order. Our product owner had an in-depth formula to calculate each item’s business value.
- Priority (or rank order): Each item had to have a unique place on the product backlog.
- Product version: Each item had to have a suggested place where it fit into the over-arching product roadmap. Defining the roadmap was more of a senior management activity, so this activity served us by aligning the product backlog items with the product roadmap.
Product Backlog Grooming
Armed with a product backlog including backlog items that had completed product backlog planning, the product owner and development team once again engaged to conduct another grooming session. The results this time were a lot more positive and productive. The goal of the product backlog grooming session, as we defined it, was to empower the development team with enough details so they could focus on execution and implementation.
The product backlog planning session had empowered the product owner to participate in meaningful discussions during the grooming sessions. By The Scrum Guide, the development team worked to refine the backlog list further and took care of providing story point estimates and selecting which items would go in which sprint. For each backlog item, the development team and product owner agreed on its acceptance criteria, estimates, and the refinement and agreement of its rank, business value, and product placement. As always, any attribute of a backlog item could be questioned, and it was the product owner’s responsibility to address those concerns.
Grade “A” Marks
Our experiment was a success. The development team was empowered and felt confident moving backlog items into sprints to move the product forward. The product owner felt empowered to drive the product backlog forward. The product owner had gained insight and support from the planning team and brought this knowledge to the development team for implementation. Because everything was stored in the product backlog, everything was transparent for all parties to see and process. The team agreed to continue to groom backlog items in this manner, and future grooming sessions proceeded smoothly.
In doing a retrospective of our experiment, we realized that we had actually implemented the very principles of hoshin kanri, a Japanese strategic planning process designed to ensure that the mission, goals, and objectives are communicated throughout an organization and implemented by everyone. The Scrum team had established a common goal, the product backlog, for all of us to focus on while involving the cross-functional team of system architects and business managers, who could align their support for the development team and product owner. We accomplished all of this while making sure we maximized our time and resources.
Who knew good product backlog hygiene could be so rewarding?
User Comments
Hi Darin,
Nice article. I appreciate your sharing it. Two comments or sort of concerns:
First, there's lots of language around maximizing everyone's time around their skills and not wasting their time. Doesn't this open the door to silo thinking and behavior within each of these "grooming teams" and across teams?
For example, I could see potentially "wasting" the time of an architect IF their involvement with the implementation team created a more robust product feature that delighted the customer with its value.
The second concern is are you leaving x-team collaboration insights on the table? Again, I wonder if the implemntation team would gain some interesting experience and skill be "being a part of" the initial backlog planning?
While I can see your need for the 2-phased meeting approach, I would personallyh work really hard to make the "team" grooming happen as much possible. And I'd want their to be some "discovery" going on in there.
Just a nuance or balance that I'm sure you're striking. Again, thanks for sharing with the community!
Bob Galen
First of all, great article. I want to share our pratice of product backlog hygiene: we divide work into tasks by category (features, requests, bugs, support, expedite). Each has their own process steps and some of them tend to be overcrowded like requests or features. We prioretize after new items added those categories and each time we drop 20% of lowest priority tasks. It gives us focus only on what is most importnat. Yet it is sad, that this is good only for product development.
Excellent share and i really liked your idea of breaking the backlog into grooming category. This makes lot of sense and would like try in my project.