You’ve been in the meeting. Someone says “we need to track this better” and fifteen minutes later the team has agreed to add four more numbers to the dashboard.
Nobody asked whether those numbers were the right numbers.
That’s the metric trap: we confuse measuring something with measuring the right thing. Let’s fix that.
A metric is a number that represents a decision
The shortest useful definition I know: a metric is a number you watch in order to make a decision.
That “in order to make a decision” part is doing a lot of work. It rules out vanity metrics — numbers that feel good but don’t change what you do. Pageviews, followers, likes: unless you’re about to make a specific choice based on them, they’re noise.
Ask yourself: If this number went up 20%, what would I do differently? If the answer is “nothing,” it’s probably not a metric worth tracking.
The three ingredients of a useful metric
Every good metric has three things:
- A numerator — what you’re counting (sessions, purchases, emails opened)
- A denominator — what you’re comparing it against (users, emails sent, time period)
- A decision — the specific choice this number informs
Notice that raw counts often fail this test. “We got 10,000 sessions this week” is not a metric until you ask 10,000 out of how many possible sessions? or compared to what?
The moment you add a denominator — a rate, a ratio, a per-unit — a count becomes a metric.
Leading vs. lagging: the timing trap
Here’s the subtler version of the same problem:
Lagging metrics measure outcomes that already happened. Revenue, churn, NPS. They’re real, they matter, but by the time the number moves, the thing that caused it happened weeks ago.
Leading metrics measure inputs or early signals. Email open rate before a campaign closes, trial signups before subscription revenue, page load time before you lose users. They’re predictive but noisier.
The mistake most teams make: they track only lagging metrics because those feel official, then wonder why they’re always reacting instead of anticipating.
A balanced set of metrics watches both: leading to steer in real time, lagging to confirm you’re heading the right direction.
The “one metric that matters” principle
When you’re starting something — a new product, campaign, or initiative — resist the urge to track ten things. Instead, identify the one metric that matters most right now (OMTM).
This isn’t because other things don’t matter. It’s because focus is how you learn fast. When you track one number, every experiment is clearly a success or failure. When you track ten, you can always find a number that went up, which means you never have to update your beliefs.
As the thing you’re building matures, the OMTM can evolve. Early-stage products care about activation. Growth-stage products care about retention. Mature products care about revenue efficiency. The metric should match the question you’re trying to answer right now.
Goodhart’s Law: the metric that eats itself
There’s a famous principle in economics: when a measure becomes a target, it ceases to be a good measure.
That’s Goodhart’s Law, and it’s the most important thing to know about metrics in practice.
The moment a metric becomes an objective — something people are evaluated on — people optimize for the metric rather than the thing it was supposed to represent.
Call center team measured on call time → agents hang up on customers.
Content team measured on article count → output triples, quality collapses.
Sales team measured on demos booked → demos with unqualified leads fill the calendar.
This doesn’t mean metrics are bad. It means every metric needs a counterbalance, a second metric that catches gaming. If you measure call resolution time, also measure customer satisfaction. If you measure content volume, also measure shares or time on page.
A practical checklist
Before you add a metric to your dashboard, run it through this:
- Does it have a denominator (is it a rate or ratio, not just a count)?
- Do I know what decision it informs?
- Is it a leading indicator, lagging, or both?
- Am I aware of how it could be gamed?
- Is there a counterbalancing metric?
If you can’t answer these cleanly, the metric probably isn’t ready to track yet, or it was never the right metric to begin with.
Metrics are a language for communicating about reality. Like any language, they can be precise or muddled, honest or misleading. The goal isn’t to track more, it’s to track the right things, and to know why.