The Proxy Problem: Why Your Best Metrics Are Standing In for Something Else

There's an old idea in economics that most leaders have felt without ever hearing the name for it. Once a measure becomes a target, it stops being a good measure. The moment people know they're being scored on something, their effort shifts toward the score itself instead of the outcome the score was built to represent.

This is the proxy problem, and it shows up everywhere in business because some metrics are standing in for something that's genuinely hard to measure directly.

What a Proxy Actually Is

A proxy is a number you can count, used to represent something you can't count as easily. Customer satisfaction is hard to measure directly, so companies track response time. Hiring quality is hard to measure directly, so companies track time-to-fill. Employee engagement is hard to measure directly, so companies run an annual survey and reduce it to a score out of ten.

None of these proxies are wrong to use. They're often the best available option. The trouble starts when the organization forgets the proxy was ever standing in for something else and starts optimizing the number as though it were the goal itself.

How the Proxy Replaces the Purpose

Consider a recruiting team measured on time-to-fill. Early on, a fast fill usually does mean a well-run hiring process. But once time-to-fill becomes the number leadership watches closely, recruiters start closing candidates faster to protect the metric. Screening steps get skipped. Offers go out before the second interview would have caught a mismatch. The number improves. The quality of who actually gets hired quietly declines, and nobody notices until the new hires start underperforming months later.

Or take a marketing team measured on cost per lead. Driving that number down looks like efficiency. But it's just as easy to hit a lower cost per lead by loosening the definition of a qualified lead, filling the funnel with names that were never going to buy. The metric improves. The pipeline gets noisier. Sales starts complaining about lead quality, and the marketing team points back to a dashboard that says everything is working.

In both cases, nobody lied and nobody cut a corner out of malice. They responded rationally to what was being watched. That's what makes the proxy problem so difficult to catch. The people gaming a metric are usually doing exactly what the incentive structure asked of them.

The Real Fix Isn't Fewer Metrics

The instinct after a story like this is to strip metrics away, on the theory that less measurement means less gaming. That instinct usually backfires. Removing the number doesn't remove the underlying question you were trying to answer. It just removes your visibility into it.

The better fix is to change your relationship with the number. Treat every metric as a conversation starter rather than a verdict. A metric that looks great deserves the same scrutiny as one that looks troubling, because an easy win is often the first sign that someone found a way to satisfy the target without solving the problem behind it.

A Simple Test for Any Metric You Rely On

Before you let a number drive a decision, run it through two questions.

First, what would this number look like if someone were actively trying to game it without doing anything wrong on paper? If you can picture a version of "success" that would leave you worse off, the metric needs a partner metric to guard against that outcome. Time-to-fill paired with 90-day retention tells a very different story than time-to-fill alone.

Second, who is close enough to the actual work to tell you what the number is missing? A dashboard can show you that a metric moved. It can't tell you why, or what got sacrificed to move it. That answer usually lives with the person doing the work, not the report summarizing it.

Neither question requires more data. Both require more judgment, which is the one input a spreadsheet can't generate on its own.

Where The Proxy Problem Leaves Leadership

Metrics aren't the enemy of good decision-making. They're one of the few tools leaders have for staying close to work they're no longer doing themselves. The risk isn't in measuring. It's in mistaking the measurement for the thing being measured, and building an entire management approach around a proxy that quietly stopped representing the truth a long time ago.

The leaders who avoid this trap aren't the ones who track the fewest numbers. They're the ones who never stop asking what a number is standing in for, and who treat every metric as something to interrogate rather than something to simply trust.