Growth loops versus funnels: draw yours and find the weak link

By Meet Patel · 2026-10-03 · 6 min read

In July 2018, Brian Balfour, Casey Winters, Kevin Kwok and Andrew Chen published an essay at Reforge titled Growth Loops are the New Funnels. Their opening claim is that “How does your product grow?” is simply the most important question a team has to be able to answer. Most teams answer it by drawing a funnel, and the funnel carries a property they pay for every month: it runs in one direction. In the authors' words, “Put more in at the top, get more out at the bottom.” When the input stops, the output stops.

A growth loop is the alternative drawing. This post explains what the authors mean by it, shows how to draw your own in five steps, and works through a hypothetical example with numbers so you can see where the weak link sits and how to find it.

What the funnel leaves out

The Reforge authors name three problems with funnels. Funnels create strategic silos, because product, channels and monetization get planned as separate topics. They create functional silos, because companies commonly organize teams by layers of the funnel, so an acquisition team and an activation team each improve their own rate and can end up working against each other. And funnels are one-directional, with no mechanism for reinvestment.

The third problem matters most for planning. A funnel report shows the conversion rate at every stage, and every one of those rates can improve while total growth stays flat, because nothing in the diagram says where the next batch of top-of-funnel traffic comes from. The diagram has an input and an output and no path between them. Whoever owns the acquisition budget owns growth by default, and that budget has to be renewed every period to hold the same result.

What a loop is, using the Pinterest example

The article defines the term this way: “Loops are closed systems where the inputs through some process generates more of an output that can be reinvested in the input.” Its main illustration is Pinterest, laid out in five steps:

1. A user signs up, or returns.
2. They activate on the product with specific, relevant content.
3. They save new content or repin existing content, which gives Pinterest quality signals.
4. Pinterest distributes the quality content to search engines.
5. A user finds the content through a search engine and signs up or returns, which is step one again.

Look at what makes this a loop. The output of one cycle, a saved pin that search engines can index, becomes the raw material for the next cycle's acquisition. No budget line appears anywhere in the diagram. The same funnel vocabulary shows up inside it (signup, activation), but the last arrow returns to the first box.

The authors also point out that loops can serve different purposes: new users, returning users, defensibility or efficiency. A loop can be viral or not, and the one requirement is that it closes.

Draw your own in five steps

Take a blank page and work through this sequence. It takes about an hour for a product you already know well, and expect most of the hour to go on step three.

1. Name the output. Write down the thing one cycle leaves behind that a stranger can encounter: a pin, a shared report, a public review, a paid invoice carrying your link. If the honest answer is “nothing”, you have a funnel, and that is a legitimate finding.
2. Write each step as a sentence with an actor and a verb. “A recipient opens the shared report” can be measured. “Referral” cannot.
3. Close the arrow. The last step must produce the same kind of person or asset that the first step started with. If it cannot, you have a funnel with a long tail.
4. Attach a rate and a time to every step. The rate says how much survives the step. The time says how long a cycle takes, which decides how quickly the loop compounds.
5. Assign an owner to every step. Steps in the middle, where product work meets marketing work, are the ones most likely to have nobody. That is the functional silo problem from the Reforge essay showing up on your own diagram.

If your product already has activation work underway, the diagram is a good place to connect it to acquisition. I have argued before that an activation metric can mislead, and a loop diagram makes the problem visible: activation only counts if it produces the output that feeds step one.

A worked example with invented numbers

Take a hypothetical 20-person SaaS company that sells reporting software to finance teams. Its loop is: a new user signs up, builds a report, shares it by link with a colleague or client, and the recipient signs up. The figures below are illustrative and belong to no real company.

Suppose one month's cohort is 1,000 signups. Of those, 40% build a report, which is 400 people. Of those 400, 25% share a report externally, which is 100 sharers. Each sharer sends it to an average of 4 recipients, so 400 recipients see it. Of those recipients, 20% sign up, which is 80 new users.

The loop multiplier is the product of the step rates: 0.4 × 0.25 × 4 × 0.2 = 0.08. Every cohort produces new signups equal to 8% of its own size. A multiplier below 1 means the loop adds to what the paid or organic top of the funnel brings and does not replace it. Summing the series (1,000 + 80 + 6.4 + and so on) gives 1,000 / (1 − 0.08), which is about 1,087 signups for every 1,000 you bring in from outside. The loop is worth roughly 8.7% extra volume.

Now raise the sharing step from 25% to 35%. The multiplier becomes 0.4 × 0.35 × 4 × 0.2 = 0.112, and the total becomes 1,000 / 0.888, about 1,126. That ten-point gain in one step is worth 39 more signups per 1,000 brought in. The multiplier has to reach 1 before the loop sustains itself with no outside input, and from 0.08 that is a 12.5-fold improvement. The arithmetic tells you how far a given loop is from carrying growth on its own, which a funnel diagram never reveals.

Finding the weak link

Because the multiplier is a product, a 10% relative improvement in any one step moves it by the same 10%. The priority goes to the step with the most headroom per week of effort, whatever its current rate. The question is which step you can move furthest in the least time.

Suppose that in the example the share action is a link buried in a menu, while the 20% recipient signup rate already sits close to what the product gets from any cold visitor. The share step is then the weak link, and the fix is cheap. The same exercise on a different product would point somewhere else, which is why the diagram has to carry your own rates.

Cycle time is the second lever. A cycle that takes 14 days completes about 6.4 times in a 90-day quarter. Cut it to 7 days and it completes about 12.9 times. At a multiplier of 0.08 the difference is small, but a loop with a multiplier of 0.7 gains a great deal from running twice as often, so measure the time before deciding the rate is the only problem.

Where the funnel still belongs

A funnel remains a sound diagnostic for a single pass through a loop: it shows where people drop within one cycle. The change is in what you call the whole system and who is accountable for it. The loop is one system with one multiplier, and a team that owns the multiplier has a reason to talk to the team that owns the step before theirs. This is also where retention as a product problem connects to growth: a returning user is an input to the next cycle as much as a new one.

When you review a growth plan, ask where the next cycle's input comes from. If the answer is a line in the budget, you are looking at a funnel and should price it as one. If the answer is something last cycle's users made, you are looking at a loop, and the work is to measure each step and fix the weakest one first.

Written by Meet Patel — startup operator and growth strategist in Dubai.

Read on themeetpatel.com