Goodhart's Law states: "When a measure becomes a target, it ceases to be a good measure."
This law, formulated by economist Charles Goodhart, explains why companies with the best metrics still fail.
What Goodhart's Law Really Means
When you tell a team "hit this number," they will hit the number. The question is: will they hit it in a way that benefits the business?
Often, the answer is no.
Real-World Examples
Example 1: Sales and Revenue
Metric: Deal count (number of deals closed) Problem: Sales reps close lots of low-value deals that churn immediately Result: Revenue per deal decreases, customer acquisition cost increases, long-term revenue drops
The team hit the target (more deals closed) but the business got worse.
Example 2: Customer Support and Response Time
Metric: Average response time (target: <2 hours) Problem: Reps send empty responses ("We're looking into this") to hit the time target Result: Customers are frustrated, issues take longer to resolve, satisfaction drops
The team hit the target (2-hour response) but the business got worse.
Example 3: Product and Daily Active Users (DAU)
Metric: DAU growth (target: 20% month-over-month) Problem: Product team adds low-quality notifications to drive engagement Result: More DAU but higher churn, lower retention, declining long-term engagement
The team hit the target (DAU growth) but the business got worse.
Example 4: Engineering and Code Coverage
Metric: Code coverage (target: 90%+) Problem: Engineers write tests that hit coverage targets without actually testing functionality Result: False sense of security, bugs slip to production, product quality declines
The team hit the target (90% coverage) but the business got worse.
Why Goodhart's Law Happens
When you optimize for a single metric, three things happen:
-
Neglect of other important factors: If you only measure deal count, you ignore deal quality. If you only measure response time, you ignore resolution quality.
-
Gaming the metric: People find ways to hit the number that don't actually improve the business. Empty support responses. Low-value deals. Meaningless code coverage.
-
Loss of context: The metric was meant to measure a proxy for business health. By optimizing for the metric itself, you disconnect from the actual goal.
How to Avoid Goodhart's Law
1. Use Multiple Metrics
Never manage a business on a single metric. Always use a balanced scorecard:
- Revenue (deal count AND deal quality)
- Customer satisfaction (response time AND resolution rate)
- Product engagement (DAU AND retention AND feature adoption)
- Code quality (code coverage AND bug escape rate)
If you optimize for one and it hurts another, you'll see it immediately.
2. Measure the Outcome, Not the Activity
- ❌ Bad: "Sales reps need to close 10 deals/month"
- ✅ Good: "Sales team needs to close $100K ARR in deals with <24-month payback period"
The second metric is harder to game because it includes quality and sustainability.
3. Build in Proxies for "Business Impact"
- ❌ Bad: "Increase customer support response time"
- ✅ Good: "Decrease resolution time AND maintain 4.5+ customer satisfaction rating"
Now if reps try to hit response time by sending empty responses, satisfaction will drop.
4. Audit the Metric Regularly
Every quarter, ask:
- Are we hitting the metric but the business is getting worse?
- Are people gaming this metric?
- Would an outsider looking at our business think we're healthy if they only saw this metric?
If the answer to any is "yes," adjust or replace the metric.
5. Align Incentives with Business Outcomes
If you reward people for hitting a bad metric, you get bad behavior. Tie compensation to:
- Revenue + profitability (not just revenue)
- Customer satisfaction + retention (not just customer count)
- Product quality + user growth (not just feature count)
The Bottom Line
Metrics are tools for understanding your business, not scorecards for judgment. The moment you treat a metric as the goal instead of the measure, you've lost sight of the actual goal.
Use leading indicators. Use multiple metrics. Audit regularly. And remember: the business outcome is what matters, not the number on the dashboard.