Set Work-in-Progress Limits for Account Research

A research board with twelve active accounts can feel productive while nothing reaches an accepted handoff. Each new request receives a little attention, and each difficult case waits for the person who started it to return. A work-in-progress limit makes that unfinished commitment visible.

The limit should define when work starts and when it genuinely leaves the system. Moving a blocked task into a separate column must not automatically make the commitment disappear.

Decide what counts as started

For a bounded account assignment, work might start when the team accepts a usable request and begins its first evidence check. It ends when the agreed output is accepted, cancelled with a reason, or formally returned because the assignment cannot proceed.

Keep the intake backlog separate. An unaccepted request is not the same as a started investigation. But a case waiting for review remains unfinished even when the researcher has stopped typing.

Microsoft's Microsoft Learn: Set Work in Progress limits describes WIP limits at board stages and notes that its product treats them as soft constraints that can be exceeded and highlighted. The operational limit below is a team agreement, not a claim that a particular tool will enforce the rule automatically.

A fictional board with hidden overload

An invented three-person team has twelve started assignments: five in active research, three awaiting review, and four blocked on clarification. The team informally counts only the five active cases and starts three more because the active column looks manageable.

The new total is fifteen unfinished commitments. The blocked work has not vanished; when answers arrive, it competes with the new assignments. The review queue also needs capacity before any of those records become useful handoffs.

For a short controlled trial, the team chooses a total started-work limit of six, including blocked and review states. This number is an illustrative hypothesis based on the team's situation, not a universal ratio per researcher. The team also sets a review-stage limit of two to expose acceptance bottlenecks.

Because fifteen cases already exist, it does not delete or relabel them to make the board compliant. It pauses new starts, reviews which questions remain needed, and works down the actual queue through acceptance, justified cancellation, or explicit return to the requester.

Define what happens when the limit is reached

First look for work that can be finished or unblocked. A researcher may help prepare an evidence comparison for a reviewer, clarify a precise missing input, or close an obsolete request with the proper owner. Starting another account should not be the automatic response to waiting.

Do not force an unsupported answer merely to free a slot. A well-documented unresolved handoff can satisfy an assignment if that was an agreed completion state. Otherwise, record the blocked decision and choose whether the work should remain committed or be formally returned.

If the team returns a task to intake, preserve its prior work and reason. Do not use that move to reset its age or conceal a slow case. The requester should know what must change before it can be accepted again.

Handle urgent work without inventing free capacity

An urgent request may justify an exception, but the exception should name what is displaced and who approves the tradeoff. Count the urgent case in total WIP. A permanent urgent lane with unlimited items defeats the limit.

For the fictional team, a critical entity check enters while six items are open. The lead pauses one lower-priority assignment with a visible revised checkpoint, records total WIP as seven during the exception, and prioritizes completing the narrow check. The board tells the truth rather than pretending the seventh item is outside the system.

Review flow and quality together

During the trial, inspect assignment age, accepted handoffs, blocked reasons, and correction burden. Do not claim improvement from a smaller active count alone. A lower limit that leaves reviewers idle while urgent work waits may need adjustment.

Look for recurring bottlenecks. If review stays full, more research starts will not fix acceptance capacity. If many cases wait for missing intake facts, improve request quality before raising the limit.

End the trial with a decision to retain, adjust, or reject the proposed limit, explaining the evidence and limitations. A useful WIP rule is a tool for finishing worthwhile work responsibly, not another target that encourages people to hide uncertainty or move unfinished tasks out of sight.

Continue with Build a Research Queue That Separates Urgent and Important Work or Create a Useful Sales Research Service-Level Agreement.