What Actually Gets Measured

Christopher McCormick, Founder & CEO, Visionary Consulting

The Reformation | Week 2 of 5 | Wisdom Wednesday

Rachel had the dashboard open before the meeting started. Jeff's team, six months after his promotion: retention flat, revenue per rep up eleven percent, engagement score a shade under company average but nothing anyone would flag. Green across the board. On paper, promoting him had been the correct call.

She also had a folder she hadn't opened in the meeting yet. Three exit interviews from his team, filed over those same six months, each one different in the details and identical in the shape: a manager who could not name what troubled the person leaving, who had never once asked what they needed, who found out about the resignation the same day everyone else did. None of that was in the dashboard. None of it was in any dashboard, because nobody had built a field for it.

"His numbers are strong," said the VP running the review, closing his laptop like the conversation was already finished. "I don't think we need to spend more time here."

Rachel did not open the folder. Not yet. But she stopped believing, in that exact moment, that the dashboard was measuring the thing it claimed to measure.

The line everyone quotes and nobody checked

Everyone quotes "What gets measured gets managed," and almost no one checks where it came from. The line is routinely attributed to Peter Drucker, a mid-20th-century management thinker whose books shaped much of modern corporate practice, yet the Drucker Institute reports no record of him ever using those exact words, and the attribution is almost certainly wrong. The phrase's precise origin is obscure, but one of its strongest early antecedents is a 1956 article by V. F. Ridgway, an organizational theorist, titled "Dysfunctional Consequences of Performance Measurements," which did not celebrate metrics at all. It warned that the effects of performance measurement were poorly understood and could be harmful.

Ridgway showed that when organizations attach rewards and penalties to a narrow set of numbers, people will manage those numbers whether or not they capture what truly matters. They will optimize the indicators that are being watched even when doing so is pointless, and sometimes in ways that damage the organization's real purpose, while reported performance appears to improve.

That is not a minor translation error; it reverses the comforting slogan. The popular version reassures executives that installing dashboards and targets is how you improve behavior. Ridgway's argument, later echoed by the economist Charles Goodhart in "Goodhart's law" (when a measure becomes a target, it ceases to be a good measure) and by W. Edwards Deming, the statistician and quality-management pioneer who helped transform Japanese manufacturing, shows that instrumenting the wrong behaviors is a reliable way to undermine the very outcomes those behaviors were meant to protect, even as every dashboard in the building insists that everything is fine.

Week 1 asked whether your promotion system rewards what your values statement claims to reward. This week asks the harder version of the same question. Your promotion system runs on your metrics. If the metric is wrong, the promotion built on top of it is wrong by construction, no matter how disciplined the review process looks from the outside.

Here's a question worth raising at your next leadership meeting: when was the last time anyone in the room asked whether a metric you've tracked for years is still measuring what you think it's measuring?

None of it shows up in the dashboard

Last month's piece on the manager whose job survived automation made this argument about a single manager, Devon, watching an AI tool do in minutes what used to take him a weekend. Once the reporting disappeared, he was left facing a senior engineer headed for burnout with no coaching plan ready, because the coaching had never been the job he was actually resourced to do. The piece named the blind spot directly: "None of that shows up in a dashboard. None of it can be automated by a tool that summarizes data faster."

Jeff's dashboard has the same blind spot, just wearing a promotion instead of an automation rollout. Revenue per rep is easy to capture, easy to defend in a meeting, and completely silent on whether the person hitting the number is coaching anyone, telling the truth about failed decisions, or building the kind of trust that makes a direct report stay through a bad quarter instead of updating a resume. The metric isn't lying. It's just answering a much smaller question than the one the VP thinks it's answering, and treating the small answer as though it settles the large one.

Try this on your own leadership dashboard: delete every metric that measures output. What's left to tell you whether your managers are actually any good?

Eight is great, until it's a federal case

In 2016, Wells Fargo disclosed that employees had opened as many as two million deposit and credit card accounts customers never authorized, a figure the bank later revised upward to approximately 3.5 million once the full review was complete. The mechanism behind it had a name employees used internally: "Eight is Great," a cross-selling quota built around the goal of eight financial products per household. Roughly 5,300 employees were terminated over the five years the practice ran, almost all of them the tellers and bankers under direct pressure to hit the number, not the executives who set it. The bank's initial 2016 settlement with regulators and the city and county of Los Angeles ran to $185 million. By 2018, cumulative penalties from the Consumer Financial Protection Bureau, the Office of the Comptroller of the Currency, a securities class action, and a fifty-state attorney general settlement had climbed past $2 billion, and the total kept growing after that.

This is Goodhart's Law running at industrial scale: when a measure becomes a target, it stops being a good measure, because the people underneath it will find the cheapest path to satisfying the number, and the cheapest path is rarely the honest one. Wells Fargo did not fail to measure sales. It measured sales with total precision and total blindness to everything the measurement was destroying to get there.

Why do you think it took a regulator to catch what "Eight is Great" was doing, and not anyone inside the building? Sit with that, then look at your own dashboard: is there a number on it that someone under enough pressure could hit without doing a single honest thing to get there?

The curve that ate Microsoft

Wells Fargo is what happens when a metric rewards the wrong behavior. Microsoft's stack ranking system is what happens when a metric is accurate and still wrong. For years, Microsoft forced managers to rate every employee against a bell curve, a fixed percentage rated poorly on every team regardless of whether that team was full of strong performers or not. The system was precise. It was also, by the company's own admission when it discontinued the practice in November 2013, a documented driver of internal hoarding and a reluctance to collaborate, because being placed on a team with other strong performers meant a mathematically worse review, curve or no curve. Lisa Brummel, then Microsoft's head of Human Resources, didn't bury the reversal in a memo. She said it in three words: "No more curve."

A metric does not have to be gamed to be dysfunctional. It only has to measure something real while ignoring that the measurement itself is changing behavior for the worse. Stack ranking told the truth about relative performance within a forced distribution and lied about everything that distribution incentivized people to stop doing for each other.

Do you know who originally designed the performance system your managers are rated on, and whether anyone has gone back to ask what it actually rewards? Somewhere in the answer is whether your best people are winning with each other or against each other.

Course completions are not the same as landing somewhere better

Reckoning's piece on what you actually owe the people whose jobs you're automating drew the sharpest version of this line the series has drawn all year. Performative reskilling announces a training platform, talks about the future of work in the abstract, and measures success by course completions. Genuine investment names, specifically, which roles are shrinking, gives people real time and real support, and measures success by how many people actually landed somewhere better, inside the company or out. One protects the organization's optics. The other protects a person's ability to plan their own life.

That distinction is not unique to reskilling. It is the distinction underneath every metric this piece has named so far. A course-completion rate is Jeff's revenue-per-rep number wearing a different department's badge: easy to report, cheap to hit, and disconnected from the outcome anyone actually cares about. If your dashboard can go green while the thing it claims to represent goes quietly wrong, you are not measuring outcomes. You are measuring activity, and calling it progress because activity is easier to screenshot.

Have you taken a look, recently, at what every metric in your own weekly report is actually a proxy for? For each one, name the human outcome it's supposed to stand in for. If you can't, the proxy has already replaced the outcome.

What MacKenzie Scott chose not to measure

Every example so far has been a warning. Here is the counter-case. July's piece on MacKenzie Scott described a philanthropist who moved more than $26 billion to over 2,700 organizations by refusing the entire infrastructure of traditional grant-making: no applications, no scored proposals, no quarterly reporting requirements that consume more of a nonprofit's time than the work itself. The piece's own words on the alternative: "complexity is not sophistication. It is often just fear with better branding," and, on what she chose to value instead of tracking it, "the work is the point. The credit is noise."

Scott did not find a better metric for grantee performance. She decided some things are better trusted than measured, and that the act of not measuring, done deliberately, is itself a form of leadership rather than an absence of rigor. That is worth sitting with, because it is the opposite instinct from everything a dashboard culture trains into people. Not every unmeasured thing is neglected. Sometimes it is the thing you trusted enough to leave alone.

None of this means stop measuring. Gallup's Q12 meta-analysis, now in its eleventh edition and drawing on over 183,000 business units, finds that teams in the top quartile of employee engagement show meaningfully lower turnover, higher profitability, and dramatically fewer safety incidents than teams in the bottom quartile, differences large enough that engagement functions as a real leading indicator rather than a soft one. The point is not that metrics are the enemy. The point is that the metrics worth keeping are the ones that survive Ridgway's question: does tracking this actually serve the purpose of the organization, or does it just look like it does from the boardroom?

Here's a question to sit with before your next planning cycle: what would you have to trust, instead of track, to find out whether it was ever worth tracking in the first place?

The audit this week actually requires

Week 1 asked you to audit who got promoted in the last two years and compare that record to what your values statement claims. This week's audit is upstream of that one: pull the five metrics that most influence a promotion, a bonus, or a performance rating in your organization, and for each one, ask Ridgway's question directly. Does tracking this serve the actual purpose it claims to serve, or does it just produce a number that is easy to defend in a meeting.

Rachel's folder has still not been opened. That's deliberate. Do you want to know why? The folder isn't the point. The dashboard that made the folder necessary is the point, and no amount of exit interviews will fix a measurement system that was never built to see the thing those interviews describe. Next week, this series turns to who gets a seat at the table before a crisis forces the issue, because a reformed metric still runs through the same narrow set of people making the calls, and that's a different failure than the one this week named.

Which brings it back to Rachel's folder. If your best people started leaving tomorrow, would your dashboard tell you why before the exit interviews did?

Where Vision Meets Reality.