2 + Fish = Moon Rock: When Your Key Performance Indicator Dashboard Has More Charts Than Purpose

When the metrics you hold so dear no longer measure anything meaningful.

Measuring the Measurers

There comes a point in every successful organization when someone walks into a meeting and says_,_ “You know what we need?”

Everyone braces.

“…more metrics.”

Not better products.

Not happier customers.

Not fewer outages.

Not faster deployments.

No.

What we really need is a dashboard with seventeen gauges, six pie charts, four heat maps, and a graph that trends confidently upward despite nobody knowing what it’s actually measuring.

Welcome to modern management.

Everything Must Be Measured

Engineers have an old saying_,_ “If you can’t measure it, you can’t improve it.”

This is generally good advice.

Unfortunately, somewhere along the way many organizations quietly replaced it with_,_ “If you can measure it, it must be important.”

This is considerably less useful.

Soon every department begins producing reports full of numbers that sound impressive:

• Story points completed
• Lines of code written
• Tickets closed
• Velocity
• Sprint burndown
• CPU utilization
• Mean clicks per user
• Number of meetings attended
• Percentage of percentages exceeding expectations

Eventually someone creates a quarterly Key Performance Indicator measuring the percentage of Key Performance Indicators that achieved their performance goals.

Congratulations.

You’ve achieved Management Inception.

Goodhart’s Law Arrives

Economist Charles Goodhart observed something in the 1970s that has since haunted virtually every corporate dashboard ever created_,_ “When a measure becomes a target, it ceases to be a good measure.”

This idea, now known as Goodhart’s Law, explains why so many well-intentioned measurement systems eventually collapse under the weight of their own incentives.

The moment people are rewarded for improving a metric, they begin optimizing the metric — not necessarily the thing it was intended to represent.

Humans aren’t broken.

They’re simply excellent optimizers.

The Software Engineering Hall of Shame

Software development has produced some legendary examples.

Lines of Code

“Developers should write more code.”

Fantastic.

Tomorrow I’ll replace:

if(x)return y;

with a beautifully documented 42-line implementation featuring decorative whitespace and inspirational comments.

Productivity increases by 4,100%.

Software quality…

Well…

Let’s not dwell on that.

Story Points

Story points were invented to estimate relative complexity.

They were never intended to compare teams.

Naturally, organizations began comparing teams.

Soon enough, “Our velocity doubled!”

Did productivity double? Maybe.

Or, perhaps every “2-point” story mysteriously became an “8.”

Code Coverage

Code coverage is useful.

Until someone decides every team must achieve 100%.

Then, you discover tests that proudly execute every line of code while asserting absolutely nothing.

The dashboard looks magnificent.

The bugs remain.

Vanity Metrics

Some metrics are particularly dangerous because they look impressive while saying almost nothing.

The analytics community even has a name for them: vanity metrics.

These are numbers that make presentations look good but don’t meaningfully inform decisions. Examples include raw page views, downloads, or social media followers when they aren’t tied to meaningful outcomes.

Imagine proudly announcing:

“We had one million API requests yesterday!”

Wonderful.

Were they successful?

No idea.

Did customers accomplish what they needed?

Unknown.

Were half of them retry storms caused by an outage?

Possibly.

But look at that graph!

Financial Reporting Loves This, Too

This phenomenon isn’t limited to IT.

Public companies have become increasingly fond of reporting carefully selected “non-GAAP” performance metrics alongside traditional financial statements.

Some of these measures are genuinely useful.

Others require a decoder ring and three accounting degrees.

You’ll encounter metrics like:

• Adjusted EBITDA
• Adjusted Adjusted EBITDA
• Normalized Operating Earnings
• Core Earnings Before Strategic Realignment
• EBITDA Before Things We’d Rather Not Discuss

None of these are necessarily inappropriate.

But, whenever companies invent increasingly customized metrics, investors should ask an important question_,_ does this number help me understand the business — or merely help the business explain the number?

Media Has Been Mocking This for Years

Popular culture has noticed.

Dilbert built an empire satirizing managers who obsess over measurements instead of outcomes.

More recently, xkcd poked fun at Goodhart’s Law with a comic whose punchline involved creating a metric for identifying bad metrics — promptly turning that into yet another target.

Reality has become self-parody.

The Dashboard Arms Race

Have you ever noticed that executive dashboards never get smaller?

Every quarter another graph appears.

Nobody removes one.

Eventually the dashboard resembles the cockpit of a commercial airliner.

Half the graphs show green arrows.

Nobody knows why.

One graph is permanently yellow.

Nobody remembers what it measures.

Everyone agrees it should probably stay.

The Key Performance Indicator Meeting

Every organization has experienced this meeting.

Manager: “We need better numbers.”

Engineer: “Which decisions are we trying to make?”

Manager: “No, no… just… more numbers.”

Six months later, someone is manually updating seventeen spreadsheets every Friday because nobody remembers who requested them, but everyone agrees they must still be important because they’ve been in the quarterly report since 2018.

Measure Outcomes, Not Activity

The best metrics answer questions like:

• Are customers happier?
• Are outages shorter?
• Are deployments safer?
• Are vulnerabilities decreasing?
• Are transactions completing successfully?

The worst metrics answer questions like:

• How many buttons did someone click?
• How many meetings occurred?
• How many Jira tickets changed color?

Activity is easy to count.

Value is harder.

Value is also what customers pay for.

A Better Approach

Metrics aren’t the enemy.

Bad metrics aren’t even the enemy.

Treating metrics as reality instead of evidence — that’s the real problem.

Use metrics as indicators.

Combine quantitative data with human judgment.

Ask whether the number still reflects the thing it was originally intended to measure.

And every so often, ask the bravest question in management_,_ “Does anyone actually use this report?”

If the answer is silence followed by several people carefully avoiding eye contact…

Congratulations.

You’ve just discovered a metric that can finally be retired.

Your dashboard — and your sanity — will thank you.

Notes

• AI / GenAI / ChatGPT / etc were not used to generate the text of this article.
• ChatGPT was used to generate the images.
• I used em dashes in my writing before the current GenAI wave was a thing. Not planning on changing now.
• Names have been changed to protect the guilty.
• None of the hostnames or users used in examples actually exist.
• Feel free to post any comments or suggestions below.