Choose metrics that lead to a decision

Name the decision before the indicator

Begin with something your team can actually decide. You might need to prioritize overdue requests, arrange coverage for a busy period or understand why work is taking longer. Write the decision in one sentence and identify who makes it. This helps you choose information that belongs in the conversation.

Then ask what evidence would change that decision. A total number of completed requests may be useful, but it may leave out how old the remaining work is. Start with the smallest set of indicators that lets the team understand both the result and the unresolved work.

Give every indicator a shared definition

An indicator needs more than a familiar name. Two people may both say “completed work” while one counts an internal handoff and the other counts delivery to the customer. Resolve that difference before comparing teams or periods. A short written definition gives people something concrete to check when a number is disputed.

Agree on how to handle cancellations, reopened requests and missing records. Keep the rule understandable to the people who enter the data. If the definition changes, make that change visible and explain whether older results can still be compared on the same basis.

• State the business question and the person who uses the answer.
• Define the event being counted and the population it belongs to.
• Describe the period, exclusions and source of the information.
• Choose a comparison that uses the same definition and time basis.
• Identify who reviews an unexpected result and when.

Put movement in context

A change in a number is a reason to investigate. Check whether the workload, mix of tasks or reporting coverage changed. A team handling more complex requests may take longer even when its work is sound. Comparing equivalent groups helps make the discussion more specific, but it still does not establish the cause by itself.

Keep the reporting period and update status visible. When information is incomplete, say what is missing. Microsoft’s dashboard guidance recommends considering the audience and giving visuals context. Use that as a presentation check after the meaning of your indicators is agreed.

A synthetic example: the backlog behind the total

Consider a fictional service team whose weekly completion total stays steady. Its oldest open requests are getting older. Looking only at completions could leave the meeting with no clear next step. Adding the age of open work reveals a question: which requests are waiting, and what is blocking them?

The team reviews the pending work, separates requests awaiting customer information and assigns a person to follow up. At the next review, it checks what changed and whether the action was appropriate. This is an invented scenario for explaining the reasoning; it does not describe a client or promise a performance improvement.

Leave the review with a next step

A useful review can end with an action, a request for more evidence or a documented decision to keep monitoring. Record the owner and the next review point. When an alert requires no response repeatedly, examine whether its definition, timing or relevance needs to change.

Periodically ask the people using the report which indicators helped them make a decision. Remove or move information that serves a different audience. Keep explanations close to the numbers so that a new team member can follow the reasoning. A dashboard becomes easier to use when its purpose remains visible as the business evolves.