Growth Loop

Growth was drawn as a funnel for a long time. Someone sees an ad, lands on the site, signs up, and pays — one pass, top to bottom. The problem with that picture is that it ends. Whoever reaches the bottom is finished, and growing again next month means pouring fresh people in at the top.

So the planning meeting came back to the same place every month. Spend $23,000 on ads this month and get 3,000 signups, and the next 3,000 cost another $23,000. Asking for five times the growth becomes asking for five times the budget. Cut the budget and arrivals drop the same month.

A growth loop is the same picture with the end attached to the beginning. If the people who reach the bottom bring the next people instead of stopping there, the 3,000 who arrived this month become part of next month's arrivals, and the second month is bigger than the first on identical ad spend. What sets the pace stops being how much money goes in and becomes what the product produces while people use it.

One turn has three parts: the input, where people or money enter; the behavior those people perform inside the product; and the path by which the result of that behavior comes back as input. Drop the third part and what's left is a funnel, not a loop.

At Pagedeck one turn runs like this.

1. A new user signs up. That's the input for this turn.
2. They write meeting notes and invite the colleagues who need to read them. Using the product is itself the act that calls the next person.
3. The invited colleague opens the link and creates an account, becoming the input for the next turn, back at step 1.

Whether the loop actually turns comes down to how many people one person brings on average. At 0.4, a hundred people bring forty, those forty bring sixteen, and it fades out. At 1.2, the same hundred bring a hundred and twenty, who bring a hundred and forty-four, and the product grows with the ads switched off. Above one or below one is the fork in the road.

How long one turn takes matters just as much. Two products both sitting at 1.2 end the year at very different sizes if one turn takes three days for one of them and three months for the other. That's the difference between a product where colleagues get invited on signup day and one where sharing only happens when the quarterly report is due.

A single blocked step stops the whole thing. That's the same story as a leaky bucket, the state where growth stalls because people leave as fast as they arrive. When retention — the share of users who stay rather than leave — is low, the people you brought in disappear before they can turn the next cycle. Hence the usual order of work: patch the holes before wiring up a new loop.

In practice a growth loop tends to take one of three shapes. What differs is the fuel.

1. Viral loop — existing users bring new users. The Pagedeck example above belongs here. It turns first in products that only become useful when several people share them.
2. Content loop — content users create pulls in search traffic. Someone publishes a meeting-notes template, it ranks in search results, and whoever finds it signs up and publishes a template of their own. It doesn't pass directly from person to person but routes through a search engine, which makes it slow. A single page takes three or four months to climb, and once it holds a position it keeps delivering people for years with no ad spend.
3. Paid loop — revenue goes back into advertising. The first two burn user behavior as fuel; this one burns money. Whether it turns depends on when what a customer pays overtakes the customer acquisition cost (CAC), the money spent to bring that customer in. Spend $23 and recover it in four months, and from month five there's money for the next person; if recovery takes twenty months, the cash to survive those months has to come from somewhere else.

Which of the three turns isn't a choice so much as a property of the product. If it's used alone and produces nothing visible outside, the first two paths are closed and only the paid one is left, which makes ad budget the growth rate. In a product where the viral path turns, fixing one invite screen gets treated with the same weight as doubling the ad budget.

"That's a funnel, not a loop. The people who finish don't bring anyone."

There's no path from the last step back to the input.

"The loop turns, but a cycle takes forever. Invites happen three weeks after signup."

The structure is right; the period is long enough that few turns fit in a year.

"Turn the ads off and arrivals go straight to zero. Nothing is looping."

A diagnosis that all the growth is being bought.

At Pagedeck the phrase comes up most often in quarterly planning. Ravi splits arrivals into people who came from ads and people who came from invites, and Grace looks at whether the second share grew against last quarter. Growing means next quarter gets bigger on the same budget; shrinking means growth stops the day the ads stop.

Mia uses the same number for roadmap order. Fixing the invite screen is work whose effect attaches to every new user from then on, so it often lands ahead of a new feature.

The most common misreading is that a loop grows without bound on its own. The end is attached to the beginning, so nothing appears to stop it. But every turn leaks somewhere, and when each person brings fewer than one other person, a hundred becomes forty, then sixteen, then nothing. Trusting it to grow by itself means an invite screen that gets a little slower each release can leave the loop standing still six months later while only the ad bill grows.

Reading it as a replacement for the funnel is the other common mistake, since the two words get used as opposites. In fact each step inside the loop is still a funnel. How many of the hundred people who saw the invite screen typed an address, how many actually sent it, and how many recipients signed up are all funnel questions. When a loop won't turn, finding the blockage means unrolling the funnel inside it.