Incentive-Driven Addiction

Definition

Sinek argues that dopamine-based incentive structures — bonuses, commissions, quarterly targets — exploit the same reward circuitry as nicotine, alcohol, or gambling: a chemical designed to reinforce survival-useful behavior instead gets triggered by arbitrary organizational targets, and when those targets are hit repeatedly without the balancing "selfless" chemicals of serotonin and oxytocin, people become "addicted to performance." The addiction is not a personal failing of employees; it is a designed property of the incentive system itself.

In the Book

Chapter 22 (titled "At the Center of All Our Problems Is Us" / "Have a Dopamine Addiction. You Earned It!") uses America Online's subscriber-acquisition incentives as its central case: one team was paid bonuses for signing up new subscribers and escalated free-trial offers from 100 hours to 250, then 700, then 1,000 free hours usable within 45 days — while a separate retention team was simultaneously trying to win back everyone the acquisition team had just signed up and lost. Sinek's diagnosis: "the leaders of AOL had effectively incentivized their people to find ways to cost the company more money," because the incentive structure rewarded each group's individual metric with no mechanism connecting either group's reward to the company's actual interest or to cooperation between them. He contrasts this with the balanced chemical loop of Paleolithic hunters, who got dopamine hits from the chase but then serotonin and oxytocin from sharing the kill with the tribe — a full loop that modern individual-metric incentive plans typically break after the dopamine stage.

Why It Matters

This distinguishes bad incentive design from bad people: a system that rewards only the individually visible, short-term metric will produce self-interested and even mutually sabotaging behavior as a predictable output, regardless of the values or character of the people inside it. It's directly exportable to any principal-agent problem — sales-vs-support conflicts, engagement-optimized software, standardized-test-driven teaching, or piece-rate manufacturing — as a question to ask before rolling out a metric: does hitting this number also require, or does it bypass, cooperation with the people the number is supposed to benefit?