Systems Thinking

Two products launch in the same month. One grows steadily for two years and then, over about six weeks, falls apart — faster than it ever grew. The other climbs quickly, flattens out around 40,000 users, and simply refuses to go higher no matter what the team ships.

Those two stories look nothing alike, but they are the same thing running in two different directions. Both are feedback loops — situations where the result of something circles back and becomes the cause of more of it. In plain terms: the output feeds the input.

There are only two kinds, and telling them apart explains most of what products do. A reinforcing loop amplifies. More users create more content, more content means better search results, better search brings more users — and round it goes, each turn bigger than the last. This is a snowball rolling downhill. Crucially, a snowball does not care which way it is pointed: the same loop that compounds your growth will compound your decline just as efficiently, which is what happened to the first product.

A balancing loop does the opposite — it pushes back toward a target. More users mean more load, more load means a slower app, a slower app means people leave, which reduces users. That is a thermostat. It is why the second product stopped at 40,000: nothing had broken, the system had simply found the ceiling where the pushback equalled the push.

When a metric that “should” keep climbing quietly flattens, the useful instinct is not to push harder. It is to go looking for what is pushing back — because something in the system is almost always doing exactly that, and no amount of extra effort at the top will out-run it.

You ship a fix for churn. Three weeks pass and the number has not moved. Was the fix wrong? Should you ship something bigger on top of it?

That question is the most expensive one in product work, and answering it well needs two ideas that sound dull and are not.

A stock is a quantity you have right now — active users, cash in the bank, unresolved bugs. A flow is what changes it: sign-ups flow in, churn flows out. The level only rises when the inflow beats the outflow, which sounds obvious until you notice how many growth plans are entirely about the inflow and never mention the drain at all.

Then the part that catches everyone: delay. Cause and effect are usually separated in time, sometimes by months. Your onboarding fix changes what happens to people signing up today — but this month’s churn number is made mostly of people who signed up long before you touched anything. The change has been made; the result has not arrived yet.

This is why so many “this isn’t working” decisions are wrong. The team ships, waits an interval that feels reasonable, sees nothing, and ships something larger on top. When the number finally moves, nobody can attribute it, and they have paid for both changes.

Estimate the delay before you start, write it down, and hold your nerve until it has passed. Most premature conclusions in product are made inside a delay nobody had measured.

Two teams get the same instruction — “reduce churn” — and the same amount of time. One rewrites onboarding copy and changes a button colour. The other changes what the sales team is paid for. A year later only one of them has moved the number, and it is not the one that worked harder.

A leverage point is a place in a system where a small, well-chosen change produces a large effect. The striking thing about leverage points is that they are rarely where effort naturally goes.

Think of a ladder. At the bottom are numbers: copy, colours, button sizes. Above them, flows — the steps in a process and the order they happen in. Above those, rules — who may do what, what requires approval. Above those, the goal — what the organisation actually rewards. And at the top, the mental model: what everyone believes without ever saying it aloud.

Effort stays roughly constant as you climb. Impact does not.

So why does everyone crowd around the bottom two rungs? Because those are the ones you can change without asking anyone. The higher rungs belong to a manager, a finance lead, a founder — which means leverage is usually a conversation rather than a ticket, and conversations are much harder to put in a sprint.

One warning comes free with this idea: every change ripples. The second-order effect — what happens next, after the obvious result — often matters more than the first. Before pulling any lever, it is worth asking out loud: “and then what happens?”