Philippines staffing research · Published · Updated
Why work-in-progress limits fail in outsourced administrative queues
A cap on open cases is not the same as a cap on work. Limits fail when they count the wrong unit, ignore review, or have no pull rule and no authority behind them.

Research findings
A work-in-progress limit is meant to stop a team from starting more than it can finish. The idea is simple: cap the items in progress so the work flows instead of piling up. In an outsourced administrative queue, the limit often fails within a month. The failure is rarely about the concept. It is about the boundary the limit draws and the authority the team has to enforce it.
The first failure is a limit set on the wrong unit. A cap on open cases counts items that are waiting for a customer, a document, or another team. Those items are not in progress, and counting them makes the limit look full when the team is idle. Measure the items actually being worked, and keep the waiting items in a separate state. The limit should describe capacity, not inventory.
The second failure is a limit without a pull rule. A cap says how many items may be active, but it does not say who may start the next one. Without a pull rule, the fastest worker starts new items while others are still finishing. The cap is exceeded by the people who are most productive, which punishes the wrong behavior. Name who pulls and when, and make the rule visible in the queue.
The third failure is a limit that the client can override. An account owner who can add an urgent item at any time turns the cap into a suggestion. The override may be legitimate, but it needs a cost and a record. If every request is urgent, the limit is not a limit. Define the exception path and the person who can authorize it.
The fourth failure is a limit with no consequence for the queue age. A team can keep the cap and let old items sit in the waiting state indefinitely. The cap looks healthy while the customer waits. Pair the limit with an age measure for each state. Then the team can see whether the cap is protecting flow or hiding a blockage.
The fifth failure is a limit that ignores the review step. In many administrative processes, work is not done until it is checked. If the cap counts only the active work and not the review, the review queue grows while the active queue looks fine. Include the review in the flow and give it its own capacity. A limit that ignores the review is measuring half the process.
Discussion
The sixth failure is a limit imposed without a say from the team. A cap changes how people work, and a cap that arrives as a rule from above is easy to work around. Explain the purpose, ask which items are hardest to finish, and adjust the number with the evidence. The team will enforce a limit it helped design. It will quietly ignore one that ignores its reality.
A useful limit starts with a measurement, not a target. Track the items in progress and the time each item spends in the state for two weeks. Then set the limit near the observed capacity, not at the theoretical maximum. The first limit should be easy to meet so the team can see the effect. Increase it only when the flow is stable at the lower number.
The limit should be paired with a definition of done. An item is not finished because a field was saved. It is finished when the evidence, the customer message, and the handoff are complete. Write the definition and use it in the daily review. A limit without a definition of done encourages people to move items out of the state rather than finish them.
Visual management helps more than a dashboard. A simple board with columns for waiting, active, review, and done lets the team see the limit and the blockage at a glance. The board should show the age of each item, not only the count. A daily walk through the board is short and surfaces the problem before it becomes a customer issue. The dashboard can follow, but the board comes first.
Implications and limits
Expect the limit to expose a constraint elsewhere. When active work is capped, the bottleneck often moves to intake, review, or a client approval. That movement is useful because it shows where the real constraint lives. Do not remove the limit to hide the new queue; fix the constraint or set a limit for it as well. A system of limits is more stable than a single cap.
Review the limit after a month with three questions. Did the active count stay at or below the cap, did the age of waiting items fall, and did the team finish more items than before? If the answer is no to all three, the limit is not working. Adjust the unit, the pull rule, or the capacity rather than adding another rule. The review should change one thing at a time.
The final test is whether the limit survives a busy week. A limit that holds only in a quiet month is not a control. Practice the exception path before the busy period and confirm who can authorize an override. Then review the overrides and their cost. A limit that is understood, enforced, and reviewed is a better tool than a target that everyone ignores.
Choosing the unit for the limit
Count the items that someone is actively working, not every open case. An item waiting for a customer reply is inventory, not work in progress. An item waiting for a review is work in progress for the reviewer. Give each state its own count and its own limit. Then the team can see where the capacity actually sits each day.
Keep the unit consistent with the process. If one case can contain several documents, decide whether the limit counts cases or documents. Counting the smaller unit gives a finer view but more administration. Counting the larger unit is simpler but can hide a heavy case. Write the choice down and use it in every report so the numbers stay comparable.
Writing a pull rule
A pull rule names who may start the next item and when. In a small team, the person who finishes an item can pull the next one from the waiting state. In a larger team, a coordinator may pull for the group at a set time. Either way, the rule should be visible and applied consistently. A pull rule turns a cap into a working system.
Record the pull in the queue so the history is visible. A short timestamp and the person who pulled are enough. Then the review can see whether the cap was respected and whether the pull matched the priority. If the priority changes often, the pull rule needs a way to reorder the waiting state. That is a separate decision from the limit itself.
Limits for review and handoff
Review is a capacity step, not an afterthought. Give it a limit based on the reviewer's available time and the complexity of the cases. A reviewer who is also handling active work will let the review queue grow. Separate the roles or reserve a block for review. The limit should reflect the time the reviewer actually has in a normal week.
Handoff needs a limit too. An item that is finished but not handed to the customer is still open from the customer's point of view. Count the handoff as a state and set a maximum age. Then the team can see whether the last step is the bottleneck. A process that is fast until the handoff is not fast for the customer.
When the client adds urgent work
Urgent work is a real part of administrative support, so the limit needs a route for it. Define what qualifies, who can authorize it, and which item is displaced. Then record the override and review it at the end of the week. An override that is never reviewed becomes the normal path. The route should protect the limit, not dissolve it.
If overrides happen every day, the limit is too low or the intake is too loose. Look at the source of the urgent requests before raising the cap. A client who sends everything as urgent may need a priority conversation. A team that cannot say no may need a decision rule. The limit is a symptom of that conversation, not a substitute for it.
Reviewing the limit
Use a short monthly review with the flow measures: active count, waiting age, review age, and completed items. Compare them with the prior month and note the one change that was made. If the numbers did not move, the change was too small or the limit is not being enforced. Choose one adjustment for the next month and write down why you chose it.
Ask the team what the limit made harder. A cap can expose a tool problem, a missing permission, or an unclear definition of done. Those are the real constraints the limit was meant to reveal. Fix one of them before changing the number. The limit is a diagnostic tool as much as a control, and it should be read that way.
Questions
Should the limit count waiting items?
No. Waiting items are inventory. Count the items someone is actively working and give each waiting state its own age measure.
Who should set the limit?
The team and the role owner together, using two weeks of observed capacity. A limit imposed without the team is easy to work around.
What if the client insists on urgent overrides?
Define what qualifies as urgent, record each override, and review the pattern weekly. If the pattern is constant, the intake or the priority rule needs a conversation.
How quickly should the limit change?
One adjustment at a time, reviewed after a month. Changing several rules at once makes it impossible to know which one worked.