KPI Examples: Why the Metrics You Track Are Secretly Destroying Your Business

In September 2016, John Stumpf, the CEO of Wells Fargo, sat before the Senate Banking Committee and tried to explain how 5,300 of his employees had opened roughly 3.5 million fake bank accounts and credit cards without customers' knowledge or consent. The number alone was staggering, but the mechanism was worse. Wells Fargo's internal culture had organized itself around a single key performance indicator: cross-selling, the number of financial products held by each customer. The company's target was eight products per household, a goal internally branded as "Going for Gr-eight." Branch managers tracked the number daily. Employees who fell short were threatened with termination. Regional leaders held conference calls where underperformers were publicly named. And so the employees did what any organism does when survival depends on a single metric: they optimized it. They opened checking accounts customers never requested. They issued credit cards that arrived in the mail unannounced. They created fake email addresses to enroll people in online banking. They transferred funds between accounts to generate activity that would look like engagement. The KPI moved in the right direction. The company was dying underneath it.

Key performance indicators don't just measure behavior. They change it. And the neuroscience of reward processing explains why: when you attach consequences to a number, the brain stops treating that number as information and starts treating it as a target, recruiting the same dopaminergic circuits that drive compulsive behavior. The most dangerous KPI isn't the wrong one. It's the right one measured in isolation, because the brain will find a way to hit it that you never anticipated. Charles Goodhart, a British economist advising the Bank of England in the 1970s, formalized this observation into what became Goodhart's Law: when a measure becomes a target, it ceases to be a good measure. Every founder building a dashboard needs to understand not just what to track, but what tracking does to the people reading it.

The British Economist Who Saw It Coming

Charles Goodhart was not a management consultant or a Silicon Valley growth hacker. He was a monetary policy economist at the Bank of England during the 1970s, a period when central banks were experimenting with targeting specific monetary aggregates to control inflation. The British government had selected a particular measure of the money supply, known as M3, as its primary policy target. If M3 grew too fast, the thinking went, inflation would follow. So the Bank of England set targets for M3 growth and built its policy apparatus around hitting them.

What happened next was instructive. Financial institutions, aware that M3 was being targeted, began restructuring their activities to move money into instruments that fell outside the M3 definition. The measured quantity stayed within target. The underlying reality, the actual flow of money through the economy, diverged from the measure entirely. Goodhart observed this in a 1975 paper and articulated the principle that would bear his name: any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes. The metric didn't fail because it was poorly chosen. It failed because the act of targeting it changed the system it was supposed to describe.

Goodhart's insight predated the internet economy by decades, but it describes the failure mode of nearly every KPI system built since. When Uber began tracking driver ratings as a key performance indicator, drivers started offering passengers mints, bottled water, and phone chargers, not because these improved the ride but because they inflated the score. When schools in Atlanta tied teacher evaluations to standardized test scores, 178 educators across 44 schools were implicated in a cheating scandal that resulted in criminal convictions. When Vietnam was under French colonial rule and officials offered a bounty per rat tail to reduce the rat population, citizens began breeding rats. The measure that was supposed to indicate rat elimination became an incentive for rat production.

The pattern is always the same. The metric starts as a mirror. It becomes a magnet. And the behavior it attracts is whatever behavior raises the number at the lowest cost, which is almost never the behavior the metric was designed to encourage.

How Dopamine Turns a Dashboard Into a Slot Machine

The neuroscience behind Goodhart's Law operates through the brain's reward prediction system, centered on dopaminergic neurons in the ventral tegmental area and the nucleus accumbens. Wolfram Schultz, a neuroscientist at the University of Cambridge, conducted foundational research in the 1990s demonstrating that these neurons don't simply fire when a reward arrives. They fire when the brain predicts a reward, and they fire more intensely when the reward is uncertain.

Schultz's experiments with monkeys showed that dopamine neurons initially responded to the juice reward itself. But after the monkeys learned that a particular cue (a light or a tone) predicted the juice, the dopamine response shifted. It fired at the cue, not the reward. The actual arrival of juice produced no additional dopamine burst. The brain had recalibrated: the metric (the cue) had become the reward, and the real outcome was neurologically invisible.

This is precisely what happens in organizations that build their culture around KPIs. The number on the dashboard becomes the cue that triggers dopamine release. Hitting the target feels like winning. Missing it feels like loss. And the actual outcome the KPI was supposed to represent, customer satisfaction, product quality, genuine growth, fades into the neurological background. The brain doesn't distinguish between hitting a sales target through excellent service and hitting it through coercion, pressure tactics, or outright fraud. The dopamine system responds to the number, not the method.

Kent Berridge, a neuroscientist at the University of Michigan, added a critical distinction in his research on "wanting" versus "liking." Berridge demonstrated that the dopamine system drives wanting, the motivational pull toward a reward, independently of liking, the actual pleasure experienced upon receiving it. This means a KPI can generate intense motivational drive (wanting to hit the number) without any corresponding satisfaction from the underlying outcome. Employees can feel urgently compelled to close deals, ship features, or open accounts while experiencing zero connection to whether those activities actually benefit the customer. The wanting system hijacks behavior. The liking system, which would normally provide feedback about whether the behavior was worthwhile, is bypassed entirely.

This is why KPI-driven cultures so often feel frantic and hollow simultaneously. The urgency is real. The meaning is absent. The dashboard is producing motivation without purpose, which is the neurological signature of compulsive behavior.

What Does a Good KPI Actually Look Like?

If every metric is vulnerable to Goodhart's Law, the question becomes not which KPIs to choose but how to structure the system around them. Andy Grove, the former CEO of Intel who would later inspire the OKR framework, understood this intuitively. In his 1983 book High Output Management, Grove introduced the concept of "paired indicators," the practice of measuring every output metric alongside a quality metric that would degrade if the output metric were gamed. If you measure the quantity of code shipped, you pair it with the number of bugs per release. If you measure sales volume, you pair it with customer retention at ninety days. If you measure accounts opened, you pair it with accounts actively used within sixty days.

The principle is architectural: no single metric should ever stand alone. The human brain, with its dopaminergic drive to optimize whatever target is most salient, will always find the shortest path to the number. Paired indicators make the shortest path the right path, because gaming one metric automatically triggers a decline in the other.

The research supports this. A 2018 study published in Management Science by researchers at Harvard Business School examined the performance effects of narrow versus broad performance measurement systems across thousands of organizations. They found that companies using a single dominant KPI showed initial performance gains followed by progressive metric manipulation and long-term degradation. Companies using balanced, multi-dimensional measurement systems showed slower initial gains but sustained performance improvement and significantly lower rates of gaming behavior. The brain needs constraints on multiple dimensions to route optimization through legitimate channels.

The Pareto principle reveals another layer: not all KPIs are equally informative. In most businesses, a small number of metrics account for the vast majority of strategic signal. The discipline isn't tracking more. It's identifying the vital few metrics that, in combination, create a picture that is expensive to fake.

Why Lagging Indicators Lie and Leading Indicators Predict

Most dashboards are graveyards of lagging indicators. Revenue is a lagging indicator. It tells you what already happened. Churn rate is a lagging indicator. By the time you measure it, the customer is already gone. Even customer satisfaction scores, when measured through periodic surveys, are lagging indicators that reflect experiences that occurred weeks or months before the data arrives.

The distinction between leading and lagging indicators isn't just a measurement technicality. It maps onto a neurological asymmetry in how the brain processes information. Daniel Kahneman's prospect theory research demonstrated that the brain evaluates information differently depending on whether it signals a future possibility or a past outcome. Future-oriented information, processed primarily through the prefrontal cortex and the anterior cingulate cortex, engages deliberative, strategic thinking. Past-oriented information, processed through the hippocampus and emotional memory circuits, tends to trigger narrative construction and hindsight reasoning.

When a founder stares at a lagging indicator like monthly revenue, the brain constructs a story about why the number looks the way it does. This activates the narrative processing network, producing a coherent explanation that may have no causal validity. When a founder stares at a leading indicator like weekly demo requests or feature activation rates within the first 48 hours, the brain shifts into prediction mode, engaging the prefrontal circuits that actually support strategic decision-making.

The practical implication is that customer lifetime value should not be your primary KPI. It should be the outcome that your primary KPIs predict. The leading indicators, activation rate, time-to-first-value, weekly usage frequency, these are the numbers that tell you where lifetime value is headed before it arrives. They give your prefrontal cortex something to work with. Lagging indicators give your narrative brain something to rationalize.