Incentives Over Intentions: Why Behaviour Follows Rewards

Why your intentions don’t run the show

Intentions feel like causes. Rewards feel like controls. But in the world we actually live in, controls win.

People behave according to what gets reinforced—by money, status, attention, safety, ease, identity, or social approval. When those reinforcers don’t match the outcome you want, behaviour doesn’t “fail.” It simply becomes consistent with the incentives you’ve built.

The hidden engine: incentive structures

An incentive structure is the environment’s “reward function.” It’s not only formal compensation. It includes:

What gets noticed (and what gets ignored)
What gets punished (explicitly or implicitly)
What becomes easy versus costly
What signals competence in a group
Which trade-offs are repeatedly tolerated

Most incentive systems are not designed. They just happen, emerging from routines, metrics, policies, and social norms.

And then everyone adjusts—sometimes subtly, sometimes dramatically—until the system’s incentives and behaviours line up.

A useful distinction: intent vs alignment

Intentions can be sincere and still produce the “wrong” behaviour. That’s because intent targets the mind, while incentives target the motion.

The core problem is misalignment:

You intend quality
The system rewards speed
The market rewards reliability of delivery, not reliability of thinking
The person closest to the cost learns to optimise the easiest route

In the long run, people don’t act according to what they “should” do. They act according to what the system makes rational.

Three ways incentives shape behaviour

1) Direct rewards and punishments

This is the obvious layer: bonuses, penalties, promotions, access, demotions.

But direct incentives are only one channel. Many of the biggest incentives are soft, delayed, or social—yet they still steer decisions.

2) The feedback timing problem

If the reward arrives late, people will optimise for the early signal. Even if the intended outcome is long-term, the actionable information becomes “what you can see now.”

This is why a system can be designed for excellence while producing mediocrity: it trains people on what is immediately rewarded.

3) Opportunity cost and friction

Incentives aren’t always “what you gain.” Often they’re “what you avoid.”

If good behaviour is expensive—time-consuming, bureaucratic, risky—then people will rationally route around it. Behaviour then tracks not only reward, but friction.

Incentive design as reality design

In these ideas, we’ve already met patterns where systems behave differently than individuals expect. Incentive structures are one of the most reliable ones.

The reason is simple: people don’t just pursue outcomes. They pursue relative advantage within constraints. The constraint set is created by incentives.

This is why “culture” talk often disappoints. Culture matters—but mostly because it changes what people expect will happen if they act a certain way. That expectation is an incentive.

The behavioural loop: learn → adapt → stabilise

Once incentives are in place, behaviour adapts. Then the new behaviour creates data. Then leaders interpret that data—often reinforcing the incentive again.

A behavioural stabilisation can occur:

People try the easiest strategy under current incentives
Results reinforce that strategy (or at least don’t disconfirm it)
The system doubles down because it appears to be “working”
Alternatives remain underdeveloped or penalised

The system becomes coherent—sometimes “perfectly.” And that’s the danger: coherence doesn’t guarantee goodness. It guarantees alignment with the objective the incentives actually encode.

Where intentions go to die: common misreads

“We told them what matters.”

Telling is not training. Without reinforcement, it’s a suggestion. Suggestions don’t change behaviour at scale.

“Performance declined, so we raised expectations.”

Expectations increase pressure, not incentives. Pressure can even produce gaming—especially when measurement is ambiguous.

“They’re capable, so they’ll figure it out.”

Capability doesn’t beat incentives. When incentives reward specific moves, people will perform those moves—even if they know better.

A practical method: rewrite the reward function

You don’t need to redesign society. You need to redesign the local incentive structure where behaviour is going sideways.

Use this sequence.

Define the behaviour you want—specifically.

Write the current reward function: what gets noticed, rewarded, and forgiven.

Map unintended incentives: what strategies are rational under current rules.

Change one lever that shifts reinforcement (reward, penalty, timing, or friction).

Run a small test, then watch the strategy people choose, not just the outcome.

The levers you actually can pull

Measurement: what’s tracked and how it’s scored
Timing: when rewards occur relative to effort
Selection: who gets opportunities under what conditions
Resource allocation: what time, attention, and budget are easiest to access
Consequences: what happens after success, failure, and near-misses
Default paths: what the system makes the default choice

You’re looking for where the system’s incentives quietly contradict the stated intention.

Tabs: how to apply this, depending on where you sit

Pick one recurring behaviour you dislike (late delivery, rushed quality, low initiative).
List the rewards and the frictions around it.
Change a single lever and test for strategy shift within one cycle.

The long view: incentives shape what future people become

Incentives don’t only affect what people do today. They shape what skills, habits, and identities become valuable.

Over time, a reward structure selects for certain internal models. People become the kind of thinkers who can exploit what’s rewarded—and ignore what isn’t.

So the question isn’t “Are people good or bad?”
It’s “What did we train them to be?”

Your actionable takeaway

Checklist0/6

Your Turn

What outcome are you trying to get—and what reward signals in your environment are most likely to produce the opposite behaviour?

Write it as: “We say we want X. We reward Y. So we get Z.”