Pull and Open-Work Limits in Administrative Work

Why it matters

In administrative work, starting many requests does not mean finishing the same number of requests. If prepared jobs pile up at a shared approval, the activity in front of employees may increase while the rate of accepted output stays the same. The pull approach starts new work when the capacity of the next step and the readiness condition of the work are visible. The open-work limit puts an upper bound on the number of requests that have started but have not yet closed through acceptance. The aim is not to move requests to an invisible place but to manage the queue created by starting more work than the finishing capacity.

The scope of open work is defined by the start and end events. If a request in preparation, in approval or waiting for information is inside the same boundary, it stays in the open-work count. Removing it from the count because it is waiting prevents the limit from showing the real load. On the other hand, requests that have not yet been started may sit in a separate intake queue. This queue is also part of the requester's total waiting. If the internal process time falls while the intake queue lengthens, the service as a whole may not have improved. The pull rule must keep visible both the open work inside and the need waiting outside.

In steady flow there is a stock–rate–time relationship among average open work, accepted closure rate and average time in process. The calculation 12 items ÷ 3 items/day = 4 days is an example of this. The 12 items here are not a snapshot of a single busy moment but a period average; the 3 items/day is the rate of correctly completed work leaving the same boundary. Counting returned documents as new output, or not converting when using hours instead of days, breaks the relationship. When inputs are continuously higher than outputs, the steady time-in-process interpretation is not valid either.

The number of the limit does not solve all problems by itself. If the authorized decision cannot be made, if missing information arrives regularly or if jobs are constantly interrupted by different priorities, open work can jam. A very low limit may starve necessary work; a very high limit hides waiting. For this reason the limit is set as a starting hypothesis and tested with completed work and waiting data. Instead of raising individual processing speed, you look at where the whole request stops. Defining the right to start new work explicitly turns the habit of everyone filling their own work list into a shared responsibility for finishing.

Formula