The Metrics You Never Chose

Look at whatever you reviewed last week. For each number on it, ask one question — not what it means, but who decided it belonged there.

In most businesses the honest answer for every single one is that it was on a screen when somebody opened the tool.

That doesn't make them wrong. Some inherited metrics are excellent. But it does mean the set has never been examined, and a set nobody assembled is a set optimized for nothing.

How a default becomes a decision

The mechanism has two stages, and the second is what makes it stick.

First, availability. A number exists on your dashboard because some vendor could compute it easily from data they already had. Email opens, deals created, tasks completed, sessions. Nothing about your business entered that decision — it was a decision about what was cheap for a software company to display.

Then repetition does the rest. The number shows up every week. It gets mentioned in a meeting, then quoted in a summary, then trended over a quarter. Within a year it has the standing of something the business decided to care about, and challenging it feels like challenging a fact rather than a choice.

Nobody remembers that the number arrived as a default. It just kept showing up until it looked like a decision.

Meanwhile the things that would actually tell you something are expensive to produce, which is exactly why they're absent. How many clients are drifting. How many decisions routed through you unnecessarily. How much promised work is past its date. Each of those requires somebody to define it, capture data nobody currently captures, and maintain it — so none of them exist, and their absence is invisible because absence always is.

The test that sorts them

Take each number and ask: what would I do differently if this moved twenty percent?

If there's no answer, the number isn't a metric. It's a fact — and you've been spending a meeting a week reviewing facts.

Run this honestly on a full dashboard and it's fairly brutal. Most founders find two or three numbers survive out of fifteen, and the survivors are usually not the ones with the most prominent placement.

What inherited metrics cost

People optimize what's measured, faithfully. Measure activity and you'll get activity — more calls, more posts, more tasks closed — regardless of whether any of it produces anything. This is the most expensive version of the problem, because the team is working hard and correctly against the target you gave them, and nobody is doing anything wrong.

False confidence, which is worse than no information. A dashboard full of green numbers describing tool activity produces a founder who believes they have visibility. A founder who believes they have visibility stops looking for the thing that's actually wrong.

Reviews become performative. When numbers don't connect to decisions, the weekly review turns into a recitation and everyone can feel it. Attendance drops, the meeting shortens, and eventually the business concludes it isn't a metrics kind of company — when what actually happened is that it was reviewing the wrong things.

The failures that matter stay invisible. Every serious problem in a founder-led business — drift, dependency, capacity, silent churn — is invisible to activity metrics by construction. You'll measure diligently right up until something breaks, then discover nothing on the dashboard was ever pointed at it.

Start from the decision, not from the data

The inversion is simple and almost nobody does it.

Instead of asking what can we measure, list the six to ten decisions you actually make on a recurring basis. Whether to take on more work. Whether to chase this deal. Whether an account needs attention. Whether to hire. Whether this month is fine.

Those decisions are stable, they're the real requirements document, and you can write them down in twenty minutes.

Then for each one, name the smallest number that would change your answer.

Frequently that number doesn't exist and can't be produced by any tool you own. That's the correct and useful outcome — you've found a real gap rather than another chart. A business that discovers it has no instrument for is this client drifting has learned something more valuable than any dashboard would have told it.

The goal isn't to watch everything. It's to know the right things to watch.

Then delete aggressively

Anything on the current dashboard that doesn't map to a decision comes off. Not archived — off, because a number that stays visible will get discussed.

The target is a short list where every entry has a decision attached. Five numbers you act on beat forty you review, and the difference isn't marginal — it's the difference between an instrument and a report.

This is the part that feels wrong and is right. Deleting a metric feels like reducing visibility. It's the opposite: it's the only way the remaining numbers get enough attention to change anything.

Accept that the good ones cost something

A number that requires capture the business doesn't currently do is a build. This is exactly where founders retreat to the easy ones, and it's worth naming the trade honestly.

One expensive number that changes a decision, against ten free ones that change nothing. That's the entire difference between telemetry and reporting.

And the capture has to be designed carefully or it won't survive. If recording it is a separate task, it'll be accurate for six weeks and patchy by month four. It has to ride along with something people are already doing — a conversation that happens in a system that records it, a stage transition that already occurs, a date that's already being set.

What a real set looks like

Three properties, and they're all structural rather than about which metrics you pick.

Every number has a decision attached. Written down, next to it. If you can't state the decision, the number comes off.

Every number has a written definition. What's included, what's excluded, and at what moment it counts. Two people using one word for two definitions is the most common reason a reporting layer stops being trusted, and it's a language problem that no tool has ever fixed.

The numbers arrive rather than wait. A dashboard is a place you have to remember to go, and remembering is the faculty that fails during a busy month. Push, never pull. If the founder has to remember to check it, it isn't telemetry — it's homework.

The decisions worth starting from

Writing down your recurring decisions sounds abstract until you try it, so here's the list most founder-led service businesses converge on. Six or seven, and they're remarkably consistent.

Can we take on more work? Needs committed load against a threshold. Almost never instrumented.

Which opportunities get my attention this week? Needs open deals with no scheduled next step, and stage age against the median.

Does any client need attention right now? Needs days since a human made contact, per account, against that account's own baseline.

Is delivery on track? Needs work items past their promised date — which requires that a promised date was recorded at all.

Should we hire? Needs the capacity number over a trend, not a feeling about how busy everyone seems.

Is this month fine? Needs the two or three numbers that actually predict the next quarter rather than describing the last one.

Notice how few of those are available from any tool by default. That's the finding, and it's the honest measure of where your visibility actually stands — not how many dashboards exist, but how many of your six recurring decisions have a number attached.

The three layers, and which one you're missing

Combine the pulse, the scoreboard and the soul and you stop reacting and start recognizing. Worth being precise about which one is actually absent.

Inherited metrics are almost entirely pulse-shaped — activity, this week, in one tool. Which is why a dashboard of them can be completely current and still leave you unable to say whether the year is going well.

What's usually missing isn't more current data. It's the scoreboard — what's working over time, in terms that connect to the business rather than to the software — and the soul, which is the part that says why any of it matters. And neither of those is ever a vendor default, because neither is cheap to compute from data a vendor happens to hold.

That's what the Telemetry pillar is actually about, and it's why more dashboards so reliably fail to produce more clarity. Volume of measurement isn't evidence of visibility. A business with fifteen dashboards can be substantially blinder than one with five numbers, because the fifteen describe software and the five describe the business.

Ninety minutes, this week

Write down your six recurring decisions. Next to each, the number that would change your answer.

Then compare that list to what's on your dashboard. The gap between the two is the honest state of your visibility, and it will tell you more than any tool evaluation.