Incentives
To understand behavior, inspect what the system rewards, punishes, makes easy, makes costly and signals as important. Then compare those forces with the outcome you actually want.
Learn it → Apply it → Save as SkillPeople adapt to rewards, costs, constraints and signals they experience, not merely to goals leaders announce.
What behavior does this system make personally sensible or easier?
Compensation, product design, education, sales, management, policy, AI evaluation and organizational systems.
Fast overview
Incentives can change behavior, but their effects depend on context. Money matters, yet so do status, recognition, effort, convenience, autonomy, career consequences, social norms and intrinsic motives. A system can unintentionally reward behavior that competes with its stated goal.
External rewards can sometimes weaken existing intrinsic motivation, but that effect is not universal. Observe the incentive as experienced, predict adaptation and check what actually happens.
Mental Model Compass
Actual behavior differs from stated priorities.
Rewards, penalties, status, effort, defaults, career risk, norms and autonomy.
What choice does the system make attractive, safe or easy?
Assuming money explains every choice or everyone responds alike.
Learn it progressively
Define success beyond the metric used to track it.
Different people face different rewards, costs, information and constraints.
Include pay, penalties, recognition, effort, convenience, autonomy, social approval and career consequences.
Ask how a sensible actor could meet the rewarded target while neglecting an unmeasured outcome.
Measure the real outcome, watch for gaming and unintended effects, then adjust as behavior changes.
Interactive Incentive Map
Examples
A commission based only on signed contracts can encourage volume while hiding poor-fit customers, discounting or later churn.
Rewards can increase engagement in some tasks, yet heavy reliance on external rewards can be counterproductive when learners already value the activity. Feedback, autonomy and competence also matter.
If a team is rewarded for one benchmark, it may improve that score without improving usefulness across real tasks.
Worked application
"Quality matters," but launch count drives reviews
Desired outcome: Useful, reliable improvements that customers keep using.
Experienced incentive: Individual performance reviews emphasize the number of features launched.
Likely adaptation: Teams can favor visible launches, postpone maintenance and accept avoidable reliability debt.
Redesign: Keep delivery speed visible, but pair it with adoption, reliability, customer outcomes and team-level measures. Protect room for engineering judgment.
Reality check: Compare behavior before and after the change. If new measures create new gaming, revise again.
Common mistakes
Status, autonomy, convenience, effort, reputation and social norms can matter.
A policy can say one thing while the promotion system or workflow signals another.
People learn the rules. A useful measure can become a target to satisfy rather than an outcome to improve.
Effects vary by task, design and existing motivation. Observe rather than assume.
When not to use it
Do not explain every behavior as a response to incentives. A person may lack skill, information, time, resources or authority. Random variation also matters. Treat incentive analysis as a testable explanation, not a claim about hidden motives.
Related models
When a proxy becomes a target, optimization can weaken its connection to the real goal.
Incentive changes alter behavior, which can create effects beyond the intended first result.
Rewards and consequences feed information back into future behavior.