When More Work Tripled the Wait

A delivery team more than doubled the work it carried at once and watched its cycle time nearly triple. It finished slightly more, and paid for it in burnout and a resignation.

05 FLOW
The condition

The same team that cleared work quickly one quarter took nearly three times as long the next, and nobody could say why.

Same team, a different quarter

One quarter, a delivery team inside a global IT organization was clearing its larger pieces of work, its epics, in under one hundred days. The next quarter, the same kind of work was taking nearly three times as long, for epics that should have run one to three months.

The delivery leader kept asking the same question. Same team. Same people. Why had everything slowed down? She was not looking for someone to blame. She genuinely could not see what had changed, and nothing in front of her could show it. The team had no flow measures, no forecast, and no control over how much work it let into the pipeline at once. When the flow collapsed, nobody saw it happen. The collapse just looked like a mystery.

A dog, a tree, and a house

Before showing the leaders their own data, we put three Lego kits on the table: a dog, a tree, and a house. Three customers. Any of them, we said, could walk in at any moment and check on their build. How would you keep all three feeling taken care of?

One of the leaders answered before anyone else: build a little of each at a time. So they did. A piece of the dog, a piece of the tree, a piece of the house, around and around. We let it run, and it took a very long time.

In the second round we challenged them to find ways to build faster and gave them room to be creative. Some teams laid their pieces out in stacked order so they could be assembled quickly. Some set the pieces on top of the build instructions so the sequence was easier to follow. In the final round we asked them to build one item at a time until it was completely done before touching the next. The total time dropped sharply, landing close to what Gerald Weinberg's well-known table on context switching predicts.

Figure 1
Three rounds at the Lego table
  1. Where it broke A little of each, in turn All three builds moving at once; it took a very long time
  2. Same switching, better moves Pieces stacked in order, laid on the instructions
  3. What changed One build at a time Finish each before starting the next; total time dropped sharply
Clever technique in round two helped at the margins. The large drop came only when the leaders stopped switching between builds.

The cost of switching lands on the leaders' desks

The leaders in the room paused after the last round. They had just watched their own instinct, keep every customer feeling served, slow the whole table down. And they recognized where that instinct showed up at work: their constant introduction of competing priorities was the root cause of the context switching and the wasted capacity on their teams.

We explained that there is no real multitasking, only switching fast or switching slow. A processor that appears to run two things at once is trading between them quickly. People do the same, and what the switching really costs is memory. A meaningful share of every person's paid time goes to remembering where they left off. One thing at a time, done well, then the next, is not a slogan. It is the arithmetic.

Glass bottles moving through a connected industrial production line.

The only thing that changed

Then we showed the delivery leader the data from her team's board. We looked back across the two quarters. In the fast quarter, the team had work in progress in the teens. In the slow quarter, it had around forty epics in flight at once. That was the only thing that changed. The team had been flooded.

The slow quarter even finished slightly more work in raw count than the fast one. On a status report it might have looked like a better quarter. But people were reporting burnout, and one person had already left for another job. Those few extra epics were bought by slowing everything else down and grinding the team out.

Figure 2
Work in flight and time to finish
Epics in flight at once
TeensFast quarter
About 40Slow quarter
Epics in flight at once rose from Teens (Fast quarter) to About 40 (Slow quarter).
Cycle time per epic
Under 100 daysFast quarter
Nearly 3x as longSlow quarter
Cycle time per epic rose from Under 100 days (Fast quarter) to Nearly 3x as long (Slow quarter).
Work in progress more than doubled between the two quarters, and cycle time for epics nearly tripled. All values are approximate.
Figure 3
What the extra starts bought
Fast quarter
  • Work in progress in the teens
  • Epics finished in under 100 days
  • Team clearing work quickly
Slow quarter
  • Around 40 epics in flight
  • Cycle time nearly tripled
  • Slightly more finished in raw count
  • Burnout reported; one person left
Raw output rose slightly. Everything else got worse, and the human cost did not appear in any measure the organization was tracking.

Putting a meter on the flow

The delivery leader's assumption was not a foolish one. The team had cleared work fast, so she believed she could pour in far more and hold the pace. Keep doing what you are doing and you keep getting what you are getting. What she had missed was that she had changed what they were doing. The pace never came from the team. It came from how little she had let them carry at once. The same blind spot sat behind the gap between what the organization promised the business each quarter and what the team could actually finish.

The strategy to close it was to give the system a flow meter: establish cycle times, make throughput visible, and connect both to work in progress, so that the next decision to add work would be made against its real cost. In the pilot that followed, simple measures of how much gets finished, how much is in progress, and how long work takes replaced manual status updates. That does not make the next slowdown impossible. What changed is that a slowdown would no longer arrive as a mystery. It would arrive as a number, early enough to act on, before it arrived as a resignation.

A team's pace belongs to the amount of work it carries, not to the team, and the bill for ignoring that arrives in people.

Share With Your Team

Choose who it is for. You send it yourself from your own email; Agileon never writes to your colleagues.

Not sure where to start?

Answer five questions about where work stalls. We send back a starting plan: where to begin and what to look at first.