Some months ago, 18F started playing with kanban as a way to manage and improve our processes. (It turns out that “DO ALL THE THINGS!” was maybe not the best motto.) Kanban is a methodology that helps you to remove inefficiencies and reduce waste by visualizing workflow. It’s also used to balance capacity and demand by implementing a “pull system” and limiting the amount of work in progress. You don’t have to change anything when you first start to use it, you just put up a board, represent your current work process, put in some work limits, and start tracking what you’re doing and learning from it.
In starting to build a board to track all of 18F’s in-flight projects, we called in 18F’s Agreements team, Josh Bailes, Matt Spencer, and Victor Valdiviezo, to answer a few questions about pieces of business in the Interagency Agreements (IAA) phase of our work. We asked them a few questions. Then we told them what a WIP limit was, and the questions started flowing the other way:
“Wait, what’s WIP?”
“WIP stands for Work In Progress.”
“So a WIP limit is…”
“A cap on the number of pieces of WIP a team can have at any given time.”
“You can do that?”
“You kinda have to do that.”
“WE NEED THAT.”
Turns out, the Agreements team was drowning in work and had been having conversations just that week about how the current pace was unsustainable. Bailes was drowning in more work than any person can handle, and Spencer was saying things to him like “You know we can’t go on like this.” Their average processing time for IAAs was sixty-five days and climbing fast. When asked what they thought a manageable number of agreements in the pipeline at a given time was, they discussed it at length and decided that a good starting point was 20. So we counted the agreements they had in process at the time they decided 20 was a sustainable pace.
They had 49.
Clearly, we needed to do something.
We talked through what they have to do to create an agreement (kanbanspeak: their “value chain”) and for how many different types of agreements they do it (kanbanspeak: their “classes of service”). Turns out there are 46 steps for multiple different purposes. And that’s not even all the work they have to do! So we said “Okay, until the work in progress is down from the ridiculous 49 to the more manageable 20, we will be starting no new agreements.”
You can imagine how this was received, but the Agreements Team stood their ground. As they have clawed toward clearing their decks enough for new agreements to enter the pipeline, they have learned an amazing amount and been able to take concrete steps toward improving their efficiency.
What they learned:
- Decreased amount of work leads to increased quality of work.
- The Agreements Team was beginning agreements without all the information needed to do so.
- You won’t drown in work if your workflow is managed.
- Bottlenecks are easier to deal with when you can see them.
- Increased accountability with a physical board for all to see led to a little extra motivation
- The team was giving less attention to responsive partners because they had to focus on the unresponsive ones.
- 18F staff was largely unaware that every IAA costs between two and five thousand dollars, including those we need to do to account for time or dollar overruns.
- Life is better when you can see your work.
Concrete steps they took:
- Started requiring the requester provide the necessary information to launch the agreement process before starting work
- Transferred responsibility to process OCSIT agreements not involving 18F to the Office of Citizen Services and Innovative Technology.
- Set expiration dates on external-signature items
- Started providing deadlines for each step, both internal and external
- Improved communications to prevent modifications and overages
- Added a system for removing agreements from the flow when the partner is excessively slow to respond
- The Agreements team reached out to the Office of General Counsel to improve mutual understanding of needs, and
- Socialized the General Counsel team into the wider 18F, to help both sides of the equation empathize with the other
A couple weeks ago, the team finally cleared the last backed-up agreement, and work began on the first new agreement to be started in several months. The cheering was both startling and touching. Turns out people had been following the team’s progress toward a manageable workload like it was the New Year’s countdown in Times Square. When the last one finally flipped, well, you’ve never seen such delighted emoji in a Slack channel in your life.
But the best part?
The team isn’t drowning anymore. They’re back in charge of the work they do, instead of the other way around.
Postscript: Two days before press time, Josh Bailes sent this 18F-wide message:
Good news. Our current median transit time for agreements is 37.5 days, meaning from kick off to signature, it takes 37.5 days for an agreement to get completed. This is a vast improvement from 60 days and due to two things: 1) the WIP limits we set. They are working. 2) The great work of Matt Spencer and [the Agreements] team.This post was originally published on the 18F blog.
Have feedback or questions? Send us an email at firstname.lastname@example.org