Your growth loop is just a drawing until you measure it

Last week I was prepping a growth loop workshop with a colleague for a client we’d just started working with. Halfway through, I worked out why these sessions land for some teams and slide right off others. The shift happens the moment we stop drawing the customer journey and start specifying how to measure it. That’s when the room goes quiet, and then it gets real.

Here’s what I think: a growth loop does nothing until you attach measurement points to it. Until then you have a nice drawing, and a drawing has never acquired a single user.

Why care? Because teams present loop diagrams to leadership, feel productive, and still can’t answer the only question that matters: is this loop actually spinning, and is it slowing down? Optimize blind and you spend real engineering weeks feeding a loop that may already be dead.

The diagram is the easy part

Every loop looks reasonable on a whiteboard. A new or existing user comes into the product, uses the core functionality, and that usage triggers an output that pulls the next user in. Take WhatsApp as a personal viral loop (the product gets better when more of the people you know are on it): I install it after an SMS invite, I connect my contacts and feel I can reach everyone I know, I message my mum, she doesn’t have the app, so she gets an SMS invite. Clean. Plausible. And plausible is exactly the trap, because a loop that looks right on a slide tells you nothing about whether it works.

Measurement turns the story into a system

The interesting thing is what happens when you put a number on every arrow. Invite sent, invite opened, signup, contacts connected, first message sent, new invite generated. Now you have a funnel you can read step by step, and you can see the conversion between each stage. The WhatsApp loop stops being a picture and becomes a sequence of rates you can watch move. A good loop? One where you can name the number for every arrow in it.

Only a measured loop can be compared and managed

This is the real payoff, and you cannot reach it from a drawing. When does the loop start losing its spin? When do you need to bolt on a second loop to give it extra fuel? Which of your loops deserves most of the optimization budget? Every one of those is a question about numbers. Without them you’re guessing, and guessing with a roadmap attached gets expensive fast.

And measuring isn’t only counting clicks. When we mapped loops with Plick, the second-hand fashion marketplace, they had a genuinely strong viral loop, but it was hard to measure and we didn’t know what actually drove it. Was it a personal viral loop, where the product gets better because your friends are on it, or a social one, where sharing raises the user’s own status (the “I was first on Clubhouse” effect, also plain word of mouth)? Reforge counts around 19 types of micro growth loop, and they behave differently, so you can’t pick the right metric until you know which one you’re holding. What the Plick team did next was the right move: they went out and asked, gathered qualitative data on user motivation, and only then built features that fed the loop and put tracking on it. Measuring the motivation came before measuring the clicks.

Now you might be thinking: the diagram is still worth doing, it aligns the team, and you can’t measure everything on day one anyway. Both true. The drawing is the necessary first step, and no, you won’t have clean tracking the morning after the workshop. So keep drawing. The point is to refuse to stop there. A loop with zero measurement points is just a hypothesis you’ve quietly decided never to test. And if all you can manage at the start is one proxy metric per loop, that still counts; one real number beats a beautiful diagram every time.

So here’s the first move, and you can do it this afternoon. Take your current loop diagram and, for every arrow, write the single number that tells you whether that step is converting. Where you can’t name the number, you’ve just found the step that’s secretly broken. (And once you have those rates, if you want to know whether they’re any good, that’s what a benchmark is for: https://benchmark.scilla.studio.)