Performance measurement does more than describe work: it directs attention, influences trade-offs and changes behaviour. Good measurement therefore pairs an outcome with leading evidence and guardrails. This guide turns the five-slide note into a practical management method, with examples, decision rules and safeguards for teams that need to convert priorities into reliable execution.
Slide 1 of 5
Every metric is also a message
When leaders repeatedly ask about one number, teams learn what receives attention. If speed is visible but quality is not, people may optimise speed. If new sales are rewarded while retention is ignored, acquisition can grow at the expense of customer value.
This is why measurement design is management design.
Start with the outcome, not the available data
Write the desired outcome in plain language before opening a dashboard. Then select three types of measure: an outcome metric that shows whether value was created, a leading indicator that reveals movement early, and a guardrail that detects unacceptable side effects.
A three-part measurement set
For customer onboarding, time-to-first-value may be the outcome. Completion of the first key action within 48 hours can be a leading indicator. Support escalation or early cancellation can be a guardrail. The set is more informative than a single activity count such as emails sent.
How metrics become distorted
Research on proxy failure explains why a measure can become less useful when strong pressure is attached to it. People learn how to improve the proxy, sometimes without improving the underlying goal. Warning signs include sudden gains with no customer benefit, local optimisation, manual exceptions, data disputes and work shifted outside the measured boundary.
Use a metric review, not only a performance review
- Restate the outcome the metric is meant to represent.
- Ask what behaviour the metric encourages.
- Inspect who or what is invisible.
- Compare the number with qualitative evidence.
- Retire or redesign measures that no longer support the decision.
Example: a support team
If agents are judged only on tickets closed, complex cases may be avoided or closed prematurely. Add resolution quality and customer recontact as guardrails, and review a small sample of cases. The goal is not a larger dashboard; it is a measurement set that supports the right decision.
Frequently asked questions
How does measurement shape management?
Metrics direct attention and influence the behaviours, trade-offs and conversations that managers reinforce.
What is a guardrail metric?
It is a measure that detects harm or quality loss while another metric is being improved.
Why is one KPI often insufficient?
A single proxy rarely captures the full outcome and can encourage local optimisation.
When should a metric be retired?
When it no longer informs a decision, no longer represents the outcome, or creates harmful behaviour.
Put it into practice
Ask what behaviour your metric encourages. Use the carousel in your next working session and record the decision, owner, evidence and review date that follow.
Download the complete carousel
Keep the five-slide framework as a PDF for your next review.
International references
- American Psychological Association — Locke’s theory of goal setting
- Gollwitzer & Sheeran — Implementation intentions and goal achievement
- Sophie Leroy — The challenge of attention residue when switching tasks
- Cambridge University Press — Proxy failure in goal-oriented systems
- Atlassian Team Playbook — Prioritize project tasks for maximum value
- Atlassian Team Playbook — Daily North Star prioritization exercise
- Atlassian Team Playbook — Redesign your calendar
- Atlassian Team Playbook — Roles and responsibilities
- Asana — Priority matrix: impact and effort
- Asana — Time blocking for focused work
- Asana — Setting and tracking OKRs
- Asana — How to prioritize tasks at work
