Metrics are the Beginning of the Conversation

Against metric fundamentalism and metric rejection alike

Kent Beck is not wrong. Not entirely, at least.
He's responding to something many of us have experienced firsthand: the moment a product stops feeling like it’s made for people and starts behaving as if it were built for a dashboard. You can almost imagine the idiot PM getting that underserved promotion by annoying you, the user, no end.

Nonsense notifications, bullshit engagement, and that sudden realisation of a thoroughly enshittified experience. When Beck warns us about enshittification, or about the tyranny of metrics, I find myself nodding violently. Yes. We’ve seen this movie. Some of us even starred in it.

So let’s start with respect. The discomfort that Beck is naming is real. And it goes without saying, the attention he has in our industry is thoroughly and richly earned.

So the following isn't disagreement, but an exercise of teasing out a bit of nuance. Maybe also a gentle and respectful roast.

My central argument is that Beck's impatience is better directed at praxis rather than at theory.

If I'm being abtuse, it's mindfully so, to saturate the arguments so that they stand out more sharply. A careless reading of Beck's article would end up in the following summarisation: Metrics did this to us. We tried measuring things. It went badly. Therefore, the measuring itself must be suspect. Better to retreat to principles, instincts, craftsmanship, and judgment. Those older, nobler tools that don’t vibrate on your wrist at inconvenient moments.

Yes, yes and yes. That's what we did. But, are we sure that’s what we set out to do?
Because in most cases, we didn’t really try metrics. Not in the way one tries to become good at something difficult. What we tried was a thinner version: random and/or cargo culted dashboards (often without ownership), haphazard targets without context, proxies promoted into judges. We asked metrics to replace thinking rather than provoke it. When the results disappointed us(as they almost inevitably do) we are at a point of declaring the whole enterprise a waste of time.

This move is understandable. It’s also a little like following a recipe, but replacing all the ingredients, and then complaining that the recipe itself is rubbish.

Or, like driving with navigation. Keeping your eyes glued to the map and away from the road is definitely gonna end up in disaster. Doing the opposite, while seemingly safe, is still ineffective.

Distance changes everything

When feedback is close, immediate, and rich, metrics are often unnecessary and sometimes absurd. A craftsperson shaping wood doesn’t need a KPI for smoothness. A developer debugging code they just wrote doesn’t need a weekly report to know something is broken. In these contexts, introducing metrics doesn’t add rigor; it strips information away, with the added bonus of being perfectly annoying.

The other extreme is an explosion of distance. Millions of users, dozens of teams. Delayed, diffused, and noisy feedback. Here, refusing to measure anything at all is reckless. You can't scale intuition, you can't calibrate the judgment across several teams of engineers. Some form of instrumentation becomes vital.

Metrics, in other words, are a compensatory technology for distance.

They exist because proximity, immediacy, and fidelity of feedback have been lost. That indeed is tragic but metrics are a way of clawing back some ways to achieve clarity.

And this is where things reliably go wrong.
Metrics are introduced to cope with distance, but then almost inevitably reinterpreted as if they had restored proximity. A proxy for feedback is mistaken for feedback itself. Numbers stop being prompts and start being verdicts. Conversations close instead of opening. Optimization becomes local, relentless, and strangely hollow. Yada*3.

That is not a failure of measurement. It’s a failure of interpretation. (Dammit this almost feels like a religious cleric offering a vehement defence lol.)

The false comfort of extremes

Faced with this, our industry has a habit of swinging between two comforting extremes.

• On one side is metric fundamentalism: the belief that with enough dashboards, we can automate judgment away entirely.

• On the other is metric rejection: the belief that the very act of measuring is what corrupts us, and that wisdom lives safely on the other side of numbers.

Both positions are oddly impatient.
Principles without operational anchors dissolve into vibes. They sound good, feel right, and quietly drift when pressure arrives. Metrics without principles harden into weapons. They look objective, feel decisive, and slowly eat the thing they were meant to protect.

The harder, less fashionable work lies in holding the two together without letting either pretend to be sufficient on its own.

What metrics are actually for

Used well, metrics are not answers. They are invitations.
They exist to surface tensions, provoke questions, and force trade-offs into the open. A metric that ends a debate has already failed. A metric that starts one, even an uncomfortable one, is doing its job.
This requires patience, governance, and admitting that numbers are evidence, not authority. And it requires becoming a student of the thing, rather than trying it briefly, badly, and then declaring it corrupt.
That work is slower than declaring allegiance to one camp or the other. It doesn’t fit neatly into slogans. And it doesn’t let us off the hook.
Which may be why we’re so tempted to skip it.

A boring responsibility

If you’re reading this and nodding along, chances are you’re in a position of influence. You’re probably reasonably well paid. You probably oscillate like most of us between confidence and a low-grade, ever-present imposter syndrome. None of that disqualifies you. In fact, it’s part of the terrain.
One of the highest-leverage things leaders can do is surprisingly unglamorous: become good stewards of value.

That means being the person who gently slows the room down when a metric is being treated as a verdict rather than a prompt. It means asking, “What question is this number actually raising?” instead of “How do we make it go up?” It means reminding teams again and again that metrics are evidence, not authority.

Sometimes it means being the buzzkill.
Not the moralistic scolder-in-chief. Not the unilateral decider. But the person who keeps reopening the conversation when it’s trying to prematurely close. The one who notices when optimization is drifting away from purpose and deftly shepherds the discussion back toward principles: What are we trying not to lose here? What would success still look like if this number went down?

This is not about rejecting metrics. It’s about refusing to let them replace thinking.
And doing this well without sounding sanctimonious, without undermining agency, without turning judgment into fiat *is the work*.

It’s subtle. It’s social. It requires taste. It rarely shows up on a dashboard. But it can be immensely rewarding. If you do it well, you can take deep pride in it because it's so woefully rare.

A note on abstraction and an open door

This is an abstract piece. That’s not because the concrete examples are missing, but because they are many, and personal. I have first-hand experiences of metrics doing real harm, and of metrics doing unmistakable good.

Those stories deserve to be told properly, with specificity and care. I've been meaning to write about them, and probably will soon.

And because this framing resists easy binaries, it should invite disagreement. Hey I'm still learning this stuff, so don't accept the authority of the written word. I'm really hoping to see some further debate. If you think metrics are more dangerous than I’ve allowed for, say so. If you think principles collapse faster than I’ve suggested, argue that case. If you believe the entire enterprise of measurement in complex human systems is irredeemable, make that argument. These are not heretical positions. They are substantive ones.

What this piece pushes back against is impatience with complexity, with apprenticeship, with the slow work of stewardship. If there is a better way to hold metrics and judgment together without letting either dominate, that conversation is worth having in the open.