Your Incentives Will Beat Your Values Every Time

Every company has a values page. Collaboration. Integrity. Customer obsession. Long-term thinking. It's on the website, the careers page, maybe a wall in the lobby. And in most companies, it has almost nothing to do with how people actually behave.

If you want to know what a company truly values, don't read the page. Look at three things: who gets promoted, how people get paid, and what behavior gets quietly tolerated. That is the real values statement. Everything else is aspiration.

People are pattern-matchers, and they're good at it. They watch what gets rewarded and what gets punished, and they adjust. Not because they're cynical, but because they're paying attention. Say you value collaboration and then rank, rate, and bonus people as individuals, and they will compete, because that is what pays. Say quality matters and then celebrate whoever shipped fastest, and they'll ship fast and cut the corners. Say you're customer-first and then compensate salespeople purely on closing, and they'll close deals the customer should never have signed. The words on the wall lose that fight every single time, because the incentive is concrete and the value is a poster.

The clearest lesson in what this costs is Wells Fargo. Its stated values were the usual: customers, ethics, trust. Its actual incentive was a relentless sales quota, with pay and jobs tied to how many new accounts an employee opened. So employees opened accounts. When the CFPB brought its action in 2016, it found more than two million deposit and credit card accounts that may not have been authorized by the customers whose names were on them, opened by employees "spurred by sales targets and compensation incentives" (CFPB). Around 5,300 employees were fired for it. Those were not 5,300 unusually corrupt people who happened to work at the same bank. They were ordinary people responding, rationally, to the system they were placed inside. The values said one thing, the incentive said another, and the incentive won at a scale no all-hands speech could ever match.

When you're small, incentives matter less than you'd expect, because the founder is standing right there. People calibrate to a person. They see what earns a nod and what earns a frown, and the culture gets corrected in real time, in the room. Scale takes that away. At two hundred people the founder can't be everywhere, and the incentive system quietly becomes the manager who is. It is always on, it touches every tradeoff and every late-night judgment call, and whatever it rewards, it rewards consistently, to everyone, forever. If you scaled the headcount but never redesigned the incentives, you handed an always-on manager a set of instructions you wrote by accident and told it to run the place.

So the incentives matter enormously. The trouble is that most of what we reward runs through numbers, and numbers are far trickier than they look.

Start with the thing every leader half-knows and still gets caught by. The moment a measure becomes a target, people begin optimizing the measure instead of the thing it was supposed to represent. Economists call it Goodhart's Law, and it is nearly inescapable. Reward activity and you get busywork that photographs well. Reward a number and you get the number, produced by whatever means the number failed to forbid. Wells Fargo's "accounts opened" was a fine proxy for "customers we are serving well," right up until it became the target, at which point it stopped measuring anything real at all.

There's a quieter version of this that does even more damage, because it never looks like cheating. Every metric is a stand-in for something you actually care about. Revenue stands in for value delivered. Velocity stands in for progress. Utilization stands in for productivity. Over time, as a company grows and the metric gets handed down through layers, people forget what it was ever a proxy for. They serve the number sincerely, diligently, and completely, and lose the thing it was pointing at. The team hits every velocity target and ships a product nobody wanted. Nobody gamed anything. They served the number perfectly and lost the thing it was meant to stand for.

Which raises a question about the goals themselves, and a piece of received wisdom worth challenging. We've all been taught that good goals are SMART: specific, measurable, and the rest. And in the right setting, they are. But the research is clear that the right setting is narrower than we pretend. Specific, challenging performance goals work well when the path is known and the environment is stable, when the task is mostly a matter of effort and execution. On work that is complex, novel, or changing, they backfire. Gerard Seijts and Gary Latham found that setting a hard performance target early in learning actually hurts performance, because chasing the number consumes the very attention people need to figure out how to do the thing at all. In those conditions, learning goals win: aim to discover a set of new strategies, not to hit a number you don't yet know how to reach. In their studies, people given learning goals achieved nearly twice the market share of those handed performance targets (Seijts and Latham). And when specific goals get over-prescribed, the side effects are systematic: a narrowed focus that ignores everything the goal didn't name, less learning, and more cutting of ethical corners (Goals Gone Wild). Here is where I'd push the point further than the usual advice does. The standard version says to use learning goals early and switch to performance goals once people have figured the work out. That switch assumes the work eventually holds still. For most of what a growing company does, it never does. If you expect your market to keep shifting and your product to keep needing improvement, then the stable, solved conditions that SMART goals require don't arrive and then stay put. They barely arrive at all. So this isn't a phase you grow out of on the way to tidy quarterly targets. For the majority of the work that actually matters, learning is the permanent condition, and goals built for a finished, predictable world are the wrong tool more or less for good.

Then there's money, where the instinct is to reach for a bigger number. Money is a real motivator, and pretending otherwise is its own kind of naive. But it is a blunt tool, and for a lot of work it's the wrong one. When Dan Ariely and his colleagues tested this for the Federal Reserve, larger financial stakes improved performance on simple, mechanical tasks and made it worse on tasks that required thinking. Across their experiments, higher incentives led to worse performance on eight of the nine tasks they studied (Large Stakes and Big Mistakes). The harder and more cognitive the work, the more a big bonus can tip from motivation into pressure, and pressure makes people worse at exactly the creative, judgment-heavy work you're paying a premium for.

None of that means money doesn't matter. It means money is one lever, not the lever, and leaning on it alone leaves most of your motivational toolkit on the floor. People also work hard for autonomy, for the chance to get better at something that matters, and for the sense that the work adds up to something, which is the whole argument of Daniel Pink's Drive. They work hard when they can see their own progress, when good work gets noticed by someone whose opinion they respect, when they feel they belong to a team that would notice if they let up. The best incentives don't just pay people to produce the thing. They make people want to produce it well.

Put it together and the job gets clearer, even if it isn't easy. Design your incentives and your measures on purpose, because if you don't, you are still incentivizing something, just by accident. Keep the "why" attached to every metric, and kill or rewrite the ones whose purpose no one can name anymore. Match the goal to the terrain, and be honest about how little of your terrain ever really settles. Default to learning goals, because in a company that keeps changing, the figuring-out doesn't end. Reserve hard performance targets for the truly stable, repeatable corners of the business, which are almost always a smaller share of the work than they feel like. Use the whole toolbox, not just comp. And be most careful about what you reward and tolerate in your best people, because that is the loudest signal you send about what actually matters here.

Culture is not what you write down. It's what you reward, what you measure, what you promote, and what you refuse to tolerate. Those things will teach your people what you truly believe far faster, and far more truthfully, than any values exercise ever could. So if you want to know what your company really stands for, don't reread the deck. Look at the last ten people who got promoted and ask what they had in common. That is your real culture. The only question is whether it's the one you meant to build.