The 6 Principles of Incentive Design

There are few forces in the world as powerful and ubiquitous as incentives. They govern everything from our daily interactions and decisions to our broader organizations and societies.

Charlie Munger, true to his reputation for pithy one-liners, said it best:

“Show me the incentive and I will show you the outcome.”

Incentives appear to be uniquely binary:

• Thoughtfully-designed incentives are likely to create wonderful outcomes.
• Poorly-designed incentives are likely to create terrible outcomes.

Surprisingly, despite their importance, incentives are rarely studied in schools, business programs, or organizations.

As a result, humans remain astonishingly bad at establishing appropriate incentives. We consistently create systems that invite manipulation and open the door to unintended consequences. More often than not, we fall into the poorly-designed camp and find ourselves scrambling for answers and quick fix solutions.

In today’s piece, I will walk through some of the common flaws in incentive design and propose a core set of principles for establishing thoughtful incentives that create desired outcomes.

An Introduction to Incentives

Let’s start with a basic definition of incentives:

Incentives are anything that motivates, inspires, or drives an individual to act in a specific manner.

They come in two forms:

• Intrinsic: Internal—created by self-interest or desire.
• Extrinsic: External—created by outside factors, typically a reward (positive incentive) or punishment (negative incentive).

Much of my writing in this newsletter and on Twitter is loosely related to intrinsic incentives—our motivations, desires, and pursuits.

But today, I'll be focusing specifically on extrinsic incentives.

In a simplified general model, extrinsic incentives involve two key components:

1. Measure: The metric that the individual or group will be judged upon. The measure can be quantitative (KPIs, metrics) or qualitative.
2. Target: The level of the measure at which a reward or punishment will be initiated. The target can be specific (you receive your incentive if the KPI hits X level) or general (you receive your incentive if your manager is satisfied with your work).

But there's a problem: This simple model of incentives—which should feel familiar—often leads to undesirable outcomes and unintended consequences.

Enter stage right, the "villain" of today's newsletter: Goodhart's Law.

Goodhart's Law

Goodhart’s Law is simple:

When a measure becomes a target, it ceases to be a good measure. If a measure of performance becomes a stated goal, humans tend to optimize for it, regardless of any associated consequences.

Goodhart’s Law is named after British economist Charles Goodhart, who referenced the concept in a 1975 article on British monetary policy.

“Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.”

But the concept was popularized by anthropologist Marilyn Strathern. In a 1997 paper, she generalized the thinking and called it Goodhart’s Law.

“When a measure becomes a target, it ceases to be a good measure.”

It quickly became a mental model with considerable practical relevance—a phenomenon that describes (and accurately predicts) the failure of simplistic incentive schemes.

Let's look at a few examples and use them to build a framework for where incentives go awry.

Indian Cobras & Soviet Nails

There were simply too many cobras in India.

The British viewed cobra heads as a clean measure for cobra elimination, so it gave the population an incentive to deliver cobra heads.

The result? Locals gamed the system, breeding cobras to earn the bounties. An incentive designed to reduce the cobra population actually increased it.

The Soviet Union needed to produce more nails to fuel their military-industrial complex.

First, Soviet leadership established incentives based on the number of nails produced. The result? The factories produced thousands of tiny nails.

Next, the leadership adjusted the incentives to be based on the weight of nails produced. The result? The factories produced a few massive nails.

In both cases, the nails were useless.

Amazon "Hire-to-Fire"

Amazon believed employee turnover was a driver of long-term business success.

From the early days, Jeff Bezos had created a culture where the bottom 10% of performers would be scrubbed in order to continuously upgrade the talent level of the organization.

To drive healthy employee turnover rates at scale, it gave its managers a target rate for annual turnover.

The result? Countless articles about a “hire-to-fire” practice emerged. Managers had allegedly hired employees they planned to fire in order to meet their turnover targets.

Wells Fargo Account Openings

Senior leadership of the bank viewed new account openings as an easy way to track business growth.

To drive new account openings, it gave its junior employees account opening targets. Employees would be pushed to exceed these targets to receive their bonuses (or risk punishment if they missed them).

The result? Employees opened millions of fake accounts to hit their targets and Wells Fargo was fined billions for the obvious fraud.

A Framework for Broken Incentives

With these examples as a backdrop, we can begin to formulate a simple framework to understand where incentives go awry.

Poorly-designed incentives typically exhibit one or more of the following three characteristics:

1. McNamara Fallacy
2. Narrow Focus
3. Vanity > Quality

The McNamara Fallacy

The McNamara Fallacy is named after Robert McNamara—the U.S. Secretary of Defense from 1961-1968—whose over-reliance on quantitative metrics is widely believed to have led the U.S. astray during the Vietnam War.

The McNamara Fallacy is the flawed assumption that what can't be measured isn't important.

It is the tendency to make a decision based on observable, quantitative metrics while ignoring all others. It leads to a focus on measuring what is easy to measure versus what is actually important (i.e. what is indicative of desired outcomes).

Cobra heads, nail quantity or weight, employee turnover, and new account openings were all easy to measure quantitatively, but totally missed the bigger picture. All four incentive programs were victims of the McNamara Fallacy.

Narrow Focus

The narrow focus issue is an objective scoping issue.

If you think too narrowly about the desired outcomes of the program, you're more likely to create incentives that miss the forest for the trees.

Using the Wells Fargo example, the desired outcome was not to have more accounts opened at the bank—more appropriate would have been to define the desired outcome as growth in the number of happy, well-serviced customers.

As a rule of thumb: When in doubt, zoom out.

Vanity > Quality

The reliance on vanity metrics—like cobra heads or new account openings—that will impress superiors or the public is a recipe for disaster in incentive design.

Imagine incentivizing a social media manager on the number of followers of the account. That person is likely to start buying followers in order to hit these targets.

The vanity metric is rarely the quality metric.

The Principles of Incentive Design

With our framework for broken incentives as a foundation, let's establish the principles of thoughtful incentive design.

The six principles to consider in crafting thoughtful incentives:

1. Objectives
2. Metrics
3. Anti-Metrics
4. Stakes & Effects
5. Skin in the Game
6. Clarity & Fluidity

Let's walk through each principle.

Objectives

Deep consideration of the ultimate objectives of the incentives is critical.

What does success look like? What is the ultimate desired outcome?

This isn’t about the surface level objectives—you need to go deeper. Without upfront deep thinking on objectives, intelligent incentive design is impossible.

Always start here before moving on.

Metrics

Establish metrics that you will measure to track success.

Importantly, be sure to avoid the McNamara Fallacy—never choose metrics on the basis of what is easily measurable over what is actually meaningful.

Just because it is easy to track a specific KPI, doesn’t mean it is the right KPI to use as a measure.

Ask yourself: If you could track and measure one metric that would tell you everything you want to know about your business or organization, what would it be?

Identify a wish list of metrics with no regard for feasibility. Work backwards from there into what is possible.

Anti-Metrics

Perhaps even more important than the core metrics, establish "anti-metrics" that you measure to track unintended consequences.

I was first introduced to this idea by Julie Zhou (she calls them counter-metrics), who has done some exceptional thinking and writing on the topics of organizations and growth.

Anti-metrics force you to consider whether your incentives are fixing one problem but creating another.

In the Amazon example, an effective anti-metric may have been average tenure of newly-hired employees by cohort. If you saw this figure dipping dramatically from the start of the employee turnover incentive program, you would know something was wrong.

Anti-metrics will tell you if you're winning the battle but losing the war.

Stakes & Effects

As with all decisions, it is critical to consider and understand the stakes:

• High-Stakes = Costly Failure, Difficult to Reverse
• Low-Stakes = Cheap Failure, Easy to Reverse

If you are dealing with a program with high-stakes, you have to conduct a rigorous, second-order effects analysis (pro tip: read my thread on this).

Iterate on your metrics and anti-metrics accordingly.

Skin in the Game

To avoid principal-agent problems, the incentive designer should have skin in the game.

Never allow an incentive to be implemented where the creator participates in the pleasure of the upside, but not the pain of the downside.

Skin in the game improves outcomes.

Clarity & Fluidity

An incentive is only as effective as:

1. The clarity of its rollout.
2. The ability and willingness to adjust based on new information.

Takeaway: Create even understanding playing fields for all constituents and avoid plan continuation bias.

Closing Thoughts

When designed thoughtfully, incentives are the most powerful tool in the modern leader's toolkit.

To avoid broken incentives, be aware of the following:

1. McNamara Fallacy: The flawed assumption that what can't be measured isn't important. Leads to a focus on measuring what is easy to measure versus what is actually important.
2. Narrow Focus: Thinking too narrowly about the desired outcomes of the program, thereby creating incentives that miss the forest for the trees.
3. Vanity > Quality: The reliance on vanity metrics that will impress superiors or the public.

To create thoughtful incentives, focus on the six principles of incentive design:

1. Objectives: Identify what success looks like. Go deep on the ultimate objectives of the program.
2. Metrics: Establish metrics to track success. Never settle for what is easy to measure over what is actually meaningful.
3. Anti-Metrics: Establish anti-metrics to determine if solving one problem is creating another.
4. Stakes & Effects: Always consider the stakes (high or low) and adjust the rigor of your second-order effects analysis accordingly.
5. Skin in the Game: Avoid principal-agent problems by ensuring the incentive designer has skin in the game (i.e. participates in both the pleasure and the pain).
6. Clarity & Fluidity: Incentives are only as good as the clarity of their rollout and the ability to adjust based on new information.

I hope you find this model as productive and helpful as I have.

I’d love to hear from you about your experience with it. Please tag me on Twitter @SahilBloom as you put it into action!

‍