The adjacent possible

Why tomorrow's opportunities are hard to measure today

Every metric is a proxy.

“Retention” itself isn’t the thing you actually care about, it’s a measurable stand-in for “people got enough value to come back”. Even your most “industry standard” metrics are just proxies.

A few categories of work resist measurable metric capture:

Foundational work. The value shows up downstream and gets attributed to whatever was built on top of it.

Risk reduction. You can’t easily measure the issue that didn’t happen, the churn you prevented, or the trust you didn’t lose.

Slow moving compounding things. Brand and trust pay off over years, with feedback loops too long to A/B test.

Strategic optionality. My favorite as a product person. Making it easier to have options while moving, keeping a door open. Real value, without a number (yet).

A culture that demands a metric for everything doesn’t just risk leaving these things unmeasured, it can encourage people to stop doing them. You drift toward small, locally optimizable, easily testable work because it’s easier to answer the “what’s the impact metric?” question for them.

Goodhart’s Law. When a measure becomes a target, it stops being a good measure.

So how do you know you’ve got good outcomes without traditional metrics?

Before you build something, see if you can phrase your theory as a simple sentence like this:

“We believe doing X will lead to Y because Z. If we’re wrong, we’d expect to see A.”

It is falsifiable even when it isn’t directly quantifiable. It tells you what you’d expect to observe if you’re right and what would convince you you’re wrong.

"We believe building a shared design system (X) will help our product teams ship more consistent, higher quality UI faster (Y), because right now they rebuild the same components from scratch and each makes its own inconsistent choices (Z). We can't promise a velocity number, since we don't control what any team actually builds, we only provide the components. If we're wrong, we'd expect to see teams adopt it but report it slows them down, or quietly keep building their own components instead (A)."

Quantifiable means you can put a number on it.

Falsifiable means there’s something you could go observe that would show the idea is wrong.

A goal can have one without the other. “This will boost results by 20%” sounds good because it has a number, but if you can’t actually trace the result back to your work, it can never really be tested. It’s precise and unprovable at the same time.

Instead of hunting for one perfect metric, combine a leading proxy (the thing one step closer in the causal chain that you can observe, like adoption rate, time-to-X, or support request volume) with qualitative evidence that you treat as real data. A collection of imperfect, independent signals pointing in the same direction can be more trustworthy than a single number.

A claim like “We’d know we’re wrong if A doesn’t happen.” has no number in it, yet it’s easy to check. You just go look for the things that would have to be true if it were right.

Name your “this would mean we’re wrong” signal in advance, before you’re attached to the outcome.

Figure out the cheapest thing you could observe that would tell you you’re heading the wrong way, and look at that first. The goal isn’t to be proven right at the end. It’s to find out you’re wrong while it’s still cheap to change course.

And just accept that some calls come down to judgment plus accountability, not measurement.

A good decision can have a bad outcome, and a bad decision can get lucky. If you only ever evaluate the outcome metric, you end up rewarding luck and punishing good judgment.

So no, not every impact is measurable, and the goal isn’t to force a number onto everything. It’s to be rigorous about your causal logic, honest about which signals are proxies, willing to count qualitative evidence as evidence, and comfortable saying “this was a judgment call” when it was one, rather than inventing a metric to launder the judgment into something that looks objective.

Many of tomorrow’s opportunities aren’t measurable before they exist. If we only fund what can be measured today, we risk never creating the conditions for tomorrow's opportunities to emerge.