As Lewis Carroll once said, “If you don’t know where you are going, any road will get you there.”

I think of this wisdom every time I hear the gospelers of the “fail-fast-fail-often” creed. I suspect that their easy acceptance of failure — and rush to celebrate it — often stems from an inability to define success.

Here’s the thing: if you don’t know what success looks like, every attempt registers as a failure. (Or worse, as politicians demonstrate daily, when you don’t know what you’re doing, every attempt can be spun as a success. But that’s another rant.)

As Andy Binns and Andreas Brandstetter argue in a recent book, innovation starts with a clearly articulated goal — a North Star that lays out the firm’s strategic ambitions and guides its actions.

Start with Strategy, Not Stumbling

With the North Star approach, success isn’t measured by the number of tries but by the steps that bring you closer to that established goal. As Andrew likes to say: it’s not about how often you fail but about how much you learn — and those two things aren’t the same. Many people and firms fail repeatedly simply because they don’t learn from previous failures. Nothing to celebrate there, in my humble opinion.

The Science of Learning from Mistakes

A 2019 paper in Nature studied the role of difficulty in learning. The findings? Maximum learning happens when the optimal training accuracy is about 85%, that is, when the error rate is around 15%.

In other words, to learn effectively, you should be right five times more often than you’re wrong.

So much for failing often.

Software Is Not Everything

We need to remember that many contemporary “rules” of innovation come from Agile software development. Sure, when designing software, you don’t have time or money for extensive customer research on every feature. You run an A/B test instead, and boom — you know what users prefer.

In this case, yes, progress can be measured by the number of tested combinations. The more, the better. And the faster you reject inferior options, the better, too. You “fail” faster? Good for you.

But not all innovation works like software development.

Take drug development. The ultimate proof that a candidate drug works (is a success, in other words) comes only in a Phase III clinical trial, which costs about $1 billion to run. With failure rates of Phase III clinical trials exceeding 50%, is there any reason to celebrate a billion-dollar failure?

Or consider creative writing. Writers can’t share early drafts with future readers. They write from the beginning to the end, publish, and then — only then — find out whether they’ve written a Pulitzer contender or warehouse filler.

Lessons from the Lab

Experimental science doesn’t measure success by failures either.

A scientist starts by formulating a hypothesis — their vision of a problem. They design an experiment to test it. If confirmed (always the preferred outcome), they formulate a new, advanced hypothesis based on the new knowledge. The process repeats.

If the hypothesis proves incorrect, they return to the drawing board and try a better one.

Failures do happen. But in science, failure is a mistake in experimental design or a screwup in implementation. It’s an embarrassment you would hide from your boss and your colleagues, not something you celebrate.

A good scientist develops better hypotheses, designs experiments that bring 100% clarity, and makes few mistakes when running them. Good scientists celebrate successes, not failures.

Making Innovation Repeatable

Innovation managers can learn from the science playbook. Place hypothesis-driven experimentation at the center of your innovation process.

But before experimentation, you need a few things:

An innovation strategy. Innovation processes. Metrics. Training. Incentives.

This is what makes the innovation process predictable and repeatable — at least more so than winning the lottery.