Walmart

Enterprise operations at very large retail scale, where project throughput depended on coordination across multiple functions, and the capacity of any single one of them mattered much less.

At this scale, the space between teams is where most of the operation actually happens. No individual team is the bottleneck, which is exactly why the loss is hard to find and so easy to put down to effort.

Work stalled between functions more than inside them, and at the key handoffs, ownership, sequence, and status were inconsistent.

A stall like this is a specific failure mode, and it’s worth being precise about. A handoff without an explicit owner and an explicit next state is really a pause that nobody has been assigned to end, like a baton set down on the track between two runners who each assume the other will pick it up. The work isn’t blocked and it isn’t moving, and no one involved is doing anything wrong, so the condition persists: it produces no complaint that points back to a cause.

The time was going somewhere quite specific: into ambiguity about who held the work next.

Organizations at this scale usually respond to that by adding oversight, because oversight is the intervention that’s easy to authorize. Status meetings, escalation paths, and reporting cadence all catch a stall after it has already cost its time, so the loss becomes visible without getting any smaller, and the senior people sitting in those meetings spend their own capacity watching it. The constraint lived in how the work was arranged, somewhere more supervision couldn’t reach.

A clearer operational path, with explicit ownership and status logic attached to the work itself, where the people doing it could see it.

At every point, every unit of work carried a named owner and a defined next state. The aim was to make a stall obvious the moment it began, and that quietly changes who notices it. When the stall shows up in the work, the person holding the work sorts it out. When it only shows up in reporting, sorting it out takes someone senior, a meeting, and a week.

Nothing in the intervention made anyone work faster, and no capacity was added. The whole gain came from time the work used to spend sitting still between functions, which is the telltale signature of a coordination problem.

When completion improves with no change in effort or headcount, the capacity was already inside the organization, being spent on coordination the system should have been carrying all along.

37% increase in project completion.

Achieved without additional headcount, which in an operation this size is the whole point: the throughput was recovered from inside the organization, and none of it had to be bought.