Goodhart’s Law: When the Measure Becomes the Target (The Cobra Effect)
In the modern corporation, "Data-Driven" is the highest compliment you can pay a leader.
We build dashboards that look like the cockpit of a fighter jet. We track "Velocity," "Churn," "NPS," and "CSAT." We tie bonuses to specific numbers. We believe that if we can measure it, we can manage it.
But often, the opposite happens. The moment we place a number on a dashboard and attach a reward to it, the number improves, but the business collapses.
This phenomenon is known as Goodhart’s Law, named after British economist Charles Goodhart. In 1975, he formulated a rule for monetary policy that has become the defining curse of modern management:
"When a measure becomes a target, it ceases to be a good measure."
Why? Because humans are creative. If you give a smart employee a number to hit, they will hit it—even if they have to destroy the company to do it.
1. The Cobra Effect: A Lesson in Perverse Incentives
To understand the danger of metrics, we must look at a story from the British Rule of India.
The British government was concerned about the number of venomous cobra snakes in Delhi. To solve the problem, they applied a "Data-Driven" solution: They offered a cash bounty for every dead cobra.
At first, it worked. Dead cobras piled up. The data looked great.
But then, the enterprising locals realized something: It is easier to breed cobras than to hunt them.
People started farming cobras in their backyards to collect the bounty.
When the British government realized this, they canceled the program.
The breeders, now stuck with thousands of worthless snakes, released them into the city.
Result: The cobra population in Delhi was higher after the program than before.
This is the Cobra Effect. It happens whenever you incentivize a proxy (dead snakes) instead of the actual goal (public safety).
2. The Toxicity of Modern OKRs
We laugh at the British colonialists, but we do the exact same thing in our QBRs (Quarterly Business Reviews).
The "Bug Bounty" Trap
- The Goal: High-quality software.
- The Metric: Number of bugs fixed per week.
- The Cobra Effect: Engineers write sloppy code to create bugs, then "fix" them to hit the target. Or, they mark trivial issues as "Critical Bugs" to pad their stats.
The "Customer Support" Trap
- The Goal: Happy customers.
- The Metric: Average Handle Time (AHT) – keeping calls short.
- The Cobra Effect: Agents hang up on customers with complex problems because solving a hard problem ruins their average. The metric improves, but Churn explodes.
The "Sales Demo" Trap
- The Goal: More revenue.
- The Metric: Number of demos booked.
- The Cobra Effect: SDRs book demos with unqualified leads who have no budget, just to hit their quota. The Sales team wastes hundreds of hours talking to ghosts.
3. The Campbell’s Law Variation
Social psychologist Donald Campbell added another layer to this in 1976:
"The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor."1
The Executive Lesson:2
Metrics are useful for 3monitoring (looking at the engine temperature), but they are dangerous for incentivizing (paying the driver based on engine temperature).
As a Chief Wise Officer, you must assume that every metric you publish will be gamed. It is not a matter of "if," but "how."
4. The Artifact: The Metric Toxicity Audit
How do you use data without creating Cobras?
You must "Red Team" your metrics before you roll them out. You must ask: "If I were an evil genius, how would I exploit this rule to get a bonus while hurting the company?"
Use this audit table for every KPI in your next Quarterly Plan.
🛠️ Tool: The Metric Toxicity Audit
| The Metric | The "Evil Genius" Strategy (How to Game It) | The Counter-Measure (The Pairing) |
| "Lines of Code Written" | I will copy-paste libraries and write verbose, unreadable code. | Never use this. (Use "Features Shipped" + "Code Review Score"). |
| "Sales Calls Made" | I will call random numbers and hang up after 10 seconds. | Pair with: "Conversion Rate." (If you make many calls but convert zero, you are fired). |
| "Ticket Closure Speed" | I will close tickets without solving them ("Marked as Duplicate"). | Pair with: "Re-open Rate." (If a ticket is re-opened by the user, the closure speed doesn't count). |
| "Uptime / Availability" | I will refuse to deploy any new features because changes cause downtime. | Pair with: "Deployment Frequency." (You must stay up while moving fast). |
Summary
Data is not truth. Data is a shadow of truth.
When you confuse the map (the metric) with the territory (the reality), you get lost.
- Bad Management says: "Get the number up by any means necessary."
- Wise Management says: "Improve the reality, and use the number to verify it."
Don't be the British Governor of Delhi. Don't pay for dead snakes.
Look at the system, look at the incentives, and ask yourself: "What behavior am I actually breeding?"
Further Reading
- "Tyranny of Metrics" by Jerry Z. Muller. (The definitive book on this subject).
- "Seeing Like a State" by James C. Scott. (How governments fail when they rely on simplified data).
- "Measuring and Managing Performance in Organizations" by Robert D. Austin.
No spam, no sharing to third party. Only you and me.
Member discussion