NoEstimates at Scale in the US Federal Government
Come learn how our 3 teams, operating in a LeSS-style scaled model, experimented with a NoEstimates approach to development work and then adopted that as our way of working for a year in the US Federal government. Included in our story is a switch to Kanban, returning to Scrum, scaling down to two teams, and eventually returning to pointing work. It has been a remarkable journey that I'm excited to share!
Outline/Structure of the Experience Report
Context & Culture of Experimentation
We’re working on a program at USCIS, a US federal agency. Within our product, we have 3 scrum teams, each ranging from 6-10 team members. Some of our golden metrics include:
- Frequently deploying to production, 10+ times daily
- Code commit to production in 20-30 minutes
- Average time to acknowledge an incident ~1 minute
Our project has a phenomenal culture of experimentation. Our team has always believed that when we create an experiment with a well-formed hypothesis, and it has a good chance of failure and a good chance of success, we’ll be able to generate more information to learn from. Our conversations are constantly around asking questions like “what did we do well, what didn’t we do so well, and what did we learn?”
Conditions Leading Up to the Experiment
- Pointing had become a stale exercise within our team (your's too?). Too often our backlog grooming discussions focused on the point estimates instead of the spirit of the work.
- One of our developers was designing an experiment to bring renewed focus on the work and reduce the "Is it a 3 or 5?" debate.
- Then a difficult release resulted in a number of issues with our partner systems. We had to act fast, so we decided in a day to abandon our Scrum-based approach and switch to Kanban. And with the unrelenting stream of incoming work, we took the first step on our NoEstimates journey.
Why NoEstimates Made Sense for Us
- We had a strong, healthy culture of experimentation
- We had a historical records that demonstrated our historical average work estimate was just above 3 points
- Prior to NoEstimates, our teams:
- Monitored flow via cycle time on 3 and 6 week rolling averages;
- Managed In-Progress work with team-level WIP limits
- Learn how we evolved our grooming process to focus on controlling the scope, content and common understanding of each piece of work
- NoEstimates makes sense when you know the domain space
- NoEstimates can work in Scrum
- And why our teams made the decision after a year to switch back to pointing
- Hear about challenges our 3 teams experienced and how each adapted in its own way
- Our thoughts about when and how teams might consider experimenting with NoEstimates
- Cycle time and WIP limits are great tools in any environment
- Learn about our product team's LeSS-style approach to scaling
- How a strong culture of experimentation allowed us to embark on this journey
Ever wish you did not have to point or otherwise estimate work? Come learn about the experience our teams had over year without pointing!
Prerequisites for Attendees
- Have a thirst for experimentation.
- It is not a requirement to understand LeSS. I'll be giving a brief overview of our adaptation in the kickoff.
- An understanding of Lean concepts will help (cycle time, flow, wip limits, etc.)