Published October 12, 2026
Published October 12, 2026
Start with the decision the scorecard should support
A useful delegation scorecard is not a prettier task list. It gives an owner enough evidence to decide whether a handoff is healthy, needs a small correction, or requires a change in authority. Begin by naming those decisions. If no one can say what they will do differently after reading a metric, that metric belongs in the work system, not the executive view.
Choose a review cadence that matches the work. A daily customer queue may need a short exception view, while a monthly reporting process may need only a close-cycle review. The assistant should maintain the record; the accountable owner should interpret tradeoffs that change policy, priority, or risk.
Track outcomes before activity
Count completed outcomes, not clicks, messages, or hours in isolation. For scheduling, an outcome might be a confirmed meeting with required participants, preparation time, and correct access. For research, it might be a decision-ready brief whose material claims can be traced to sources. Activity still helps diagnose a problem, but it should not masquerade as value.
Write a short definition beside each measure. “Completed” must mean the same thing to the assistant and the owner. State the denominator, exclusions, and time window so a calm week and a high-volume week are not compared as though they were identical.
Give exceptions their own lane
An average can hide the few cases that consume most attention. Add an exception count with categories such as missing input, access failure, late owner decision, external delay, or quality rework. The point is not to blame whoever touched the task. It is to reveal where the operating design is asking people to improvise.
Keep the categories few enough to use consistently. If an exception does not fit, record a short note and decide during review whether the taxonomy needs to change. A growing “other” category is a signal that the current labels no longer describe the work.
Show workload and waiting separately
Backlog alone cannot tell you whether the assistant lacks capacity or is waiting on someone else. Split ready work from blocked work. For blocked items, record the decision owner, what is needed, and the next check time. This prevents a queue from making an assistant look overloaded when the real constraint is unresolved approval.
Use ranges rather than artificial precision. A simple capacity signal - available, near limit, or over limit - can be more honest than a percentage built from uncertain task estimates. Pair it with the next known deadline and the consequence of delay.
Add a decision column
Every scorecard review should end with explicit decisions: continue, adjust the rule, add capacity, remove work, or investigate. Record the owner and effective date. Otherwise the same concern returns each week as commentary with no operational change.
A strong decision note is brief: “Owner approved assistant replies to routine reschedule requests within the published policy beginning Monday.” It identifies changed authority without rewriting the entire procedure.
Use a compact weekly layout
One page is usually enough: three outcome measures, an exception summary, capacity status, blocked decisions, and changes agreed. Link to detailed records instead of embedding them. The scorecard is an interface to the operation, not the operation itself.
Review trends over several comparable periods before declaring improvement or decline. Volume, seasonality, new clients, or a changed definition can move a number. Annotate these context changes so later reviewers do not invent a story around a discontinuity.
Run the first review as a calibration
Ask the assistant which numbers were hard to produce and which labels required judgment. A metric that takes hours to reconstruct will not remain reliable. Prefer signals captured during ordinary work, then audit a small sample against source records.
At the end, remove any measure no one used and add only signals tied to a real decision. The result should make delegation calmer: the owner sees what matters, the assistant knows how work will be judged, and both can address exceptions without reopening every completed task.
This guide was published on October 12, 2026. For adjacent workflows, explore the Delegation Assistant blog or review the deeper evidence in Research.