Have you heard of the cobra effect? It dates back to the British occupation of India, and it goes like this.
The Brits wanted to reduce the number of cobras, so they offered a bounty for dead cobras. Bring in a dead snake, get some money.
It worked. People killed cobras and collected the bounty, and they started to rely on that additional cash in their household budgets. But as the cobra population declined, it got harder to kill cobras.
The Indians were too clever for that little problem and quickly came up with the obvious solution. Breed cobras, kill them, then turn in the carcass for the bounty.
When the government discovered what was happening, it canceled the program. The cobra breeders, now stuck with worthless snakes, released them.
The program that was designed to reduce the cobra population ended up increasing it.
This type of problem appears over and over again in business. Incentives change behavior in ways you didn’t expect or intend.
The Paradox
We can expand the cobra effect into a related concept that I’m going to call Pigweed’s Paradox.
It goes like this. An organization, department, or job is created to solve a problem. But if solving the problem eliminates the need for the organization, you’ve created an incentive to not solve the problem.
There are variations on the theme, but they all come back to the same basic problem.
The way people optimize their behavior for the incentives you give them might have unintended results.
How can this goal go haywire?
There’s a related idea called Goodhart’s Law: “When a measure becomes a target, it ceases to be a good measure.”
Imagine that you want your customer service department to be more efficient, so you start measuring the number of calls each rep handles per hour, and you rate the department on that basis.
That seems reasonable, but how will the reps respond to that incentive? They’ll cut calls short, which means customers won’t get their problems solved.
The sequence goes like this.
- A goal is converted into a proxy measurement.
- The proxy measurement is incentivized.
- The behavior produced by this incentive is not what was intended.
- The goal is not attained, and sometimes the opposite is achieved.
Along those lines, here are some common examples.
- Clicks become a proxy for engagement which incentivizes tricks to get clicks.
- Web traffic, social media followers, or email subscribers become a proxy for audience size, so you attract lots of the wrong people.
- Word count becomes a proxy for substantive articles which are now filled with fluff.
- “Top 3 position on keywords” becomes a proxy for high-value online content, but since the keywords are misaligned with your product, purchases tank.
What if the opposite is true?
Metrics and incentives are good things, but we have to think about how they will be abused.
Whenever you create a metric, a goal, or an incentive, ask a few questions.
- What wrong-headed behavior might this encourage?
- How could someone game this measurement?
- What happens if people optimize for the measurement rather than the thing the measurement is supposed to represent?
- And perhaps most importantly: What if the opposite is true?
What if the thing you’re doing to increase engagement actually reduces it? What if the campaign designed to grow your audience fills your database with people who will never buy anything? What if the productivity metric makes people less productive?
Don’t just ask whether your incentive will work. Ask what happens when it works too well, and not in the way you intended.
