Growth Loops

Originated by Reforge (Brian Balfour and others) in 2018

A model that maps how a product's output becomes the input for its own next cycle of growth — compounding value over time instead of draining a finite acquisition budget.

A funnel is a useful way to think about conversion. Users enter at the top, some percentage move through each stage, and a smaller percentage exit the bottom as customers. The metaphor captures the drop-off at each stage, which is useful for diagnosis. But it implies something that isn’t true: that growth is a one-way process, input flowing through stages until it’s consumed.

The most efficient growth systems don’t work like funnels. They work like loops. The output of one cycle becomes the input of the next. Each iteration builds on the previous one. Growth compounds rather than depletes.

That’s the core idea behind Growth Loops, a framework formalized by Reforge (the growth education company founded by Brian Balfour and others) starting around 2018.

What the Framework Actually Does

A growth loop maps the mechanism by which a product generates its own future growth. Every effective loop has the same basic structure: a user takes an action (the input), the platform does something with it (the action), something valuable is created as a result (the output), and that output generates new users or new actions that feed back into the loop as fresh inputs.

The crucial property that distinguishes a loop from a funnel is the feedback. In a funnel, outputs leave the system (a customer is acquired and then… you spend money to acquire the next one). In a loop, outputs re-enter the system and create the conditions for the next cycle to be larger than the last.

The best loops compound. Each iteration produces more than it consumed. The classic example is a content loop: users create content, the platform distributes it, new users see the content and join, they create more content, the platform has more to distribute, and more new users see it. Every cycle produces a larger audience and a larger content library, which feeds the next cycle.

Viral loops, referral loops, data loops, content loops, and community loops all share this feedback structure. They differ in what the input is, what the platform does with it, and what kind of output is generated.

The Origin

The Growth Loops framework was developed and popularized by Brian Balfour, Andrew Chen, and others at Reforge beginning around 2018, though the underlying ideas were circulating in growth practitioner communities earlier. The explicit critique that sparked the framework was directed at the AARRR funnel model: Balfour and his colleagues argued that treating growth as a linear funnel was conceptually wrong for products that had compounding growth mechanics, and that the better mental model was loops.

Andrew Chen’s research into network effects and viral growth, and his writing on topics like the “cold start problem” (how do you get a loop started when you have no existing users?), provided much of the intellectual foundation. The Reforge curriculum synthesized these ideas into a teachable framework.

The framework gained broad uptake in the growth community because it described something practitioners could already see in the data of successful products but didn’t have language for: the difference between companies where CAC kept rising and companies where growth seemed to get easier over time. Loops explained why some products could effectively stop paid acquisition and still grow, while others stalled the moment the paid channel was turned off.

How to Apply It

Start by mapping your current growth. Ask: where do new users actually come from? Be specific. Look at your acquisition data and trace each new user back to their source. If a meaningful portion of your new users are coming from existing users (referrals, shared content, word of mouth), you have the beginnings of a loop. If almost everyone is coming from paid channels or direct search, you’re probably funnel-dependent.

Then map the mechanism. What is the user action that generates the input? What does your platform do with it? What output is produced, and does that output reach new potential users? If you can draw the circle, you have a loop. If the line doesn’t close, you have a one-way channel.

Next, look for leverage points. Where in the loop can you reduce friction, increase the quality of the output, or expand the distribution of what’s produced? The Spotify Wrapped example is instructive: the design team clearly thought hard about making the shareable cards visually distinctive enough to stand out in a social feed, because the loop only works if the output is compelling enough to create genuine curiosity in people who see it.

Think about how to seed loops that haven’t started yet. This is the cold start problem. A content loop requires existing content before new users see enough to want to join. You may need to manually create the initial conditions before the loop can become self-sustaining.

Finally, measure the loop’s health. A loop is healthy if each cycle produces at least as many new inputs as the previous one. It’s compounding if it produces more. It’s declining if each cycle is producing fewer. The K-factor concept from viral marketing is essentially a loop health metric: a K-factor above 1 means the loop is compounding.

A Real Example

Spotify Wrapped is probably the most analyzed growth loop in recent consumer tech. The mechanism: Spotify users listen throughout the year (input). In December, Spotify processes listening data and generates personalized annual summaries (action). Those summaries are shared on social media by users who find them interesting and worth showing off (output). People who see the shares, curious about their own listening data, sign up for Spotify (new inputs into the loop). Each year’s Wrapped reaches a new audience, including people who weren’t yet Spotify users. The loop compounds because the audience of people who’ve seen a Wrapped card keeps growing.

The design is deliberate at every stage: the cards are visually shareable, the data is personal enough to feel worth posting, and the “your year in music” framing gives users a social reason to share (self-expression) rather than just a marketing reason (helping Spotify acquire users). The input side works because users don’t feel like they’re doing marketing. They feel like they’re expressing something about themselves.

Notion’s template ecosystem runs a content-plus-referral loop. A user creates a template (input). Notion distributes it in their template gallery, which is publicly accessible and search-indexed (action). New users discover Notion while searching for templates for their specific use case (output, which becomes a new user — a new input). That new user is already past the awareness stage and partially through activation, because they found Notion at the moment they needed it and through a concrete demonstration of its value.

Burberry’s Art of the Trench campaign was an early brand example of a UGC loop: customers submitted photos of themselves in Burberry trench coats (input), Burberry featured them on a dedicated microsite (action), other customers saw the gallery, felt inspired to submit their own photos, and younger consumers discovered the brand through the gallery rather than through traditional advertising (output that fed back as new inputs).

When the Framework Falls Short

Not every product can have a meaningful growth loop, and forcing the framing can be misleading. Some products don’t have a natural mechanism by which users generate outputs that reach new potential users. B2B tools used internally, without sharing or network effects, often acquire customers through direct sales and referrals that function more as isolated events than compounding loops. Calling that a loop doesn’t make it one.

Loops also saturate. A referral loop that works brilliantly in the early growth phase can slow as you penetrate the most connected early adopters and reach users with smaller networks or less referral motivation. The loop health metrics need to be monitored continuously; a loop that looked healthy at 100,000 users might be clearly plateauing at 10 million.

Building loops requires product investment. You can’t install a growth loop through a marketing campaign alone. The Spotify Wrapped loop required engineering, data infrastructure, and design work. The Notion template gallery required platform and indexing work. If your marketing team is thinking about loops but your product team isn’t building the infrastructure, the loop won’t close.

The framework is also descriptive rather than prescriptive. It tells you what loops look like and why they work, but it doesn’t tell you which loop to build, in what order, or how to prioritize loop development against other product work. You need judgment and context to answer those questions.

When to Use It (and When to Reach for Something Else)

Use Growth Loops when you’re trying to understand the mechanics of your best growth channels, when you want to design product features with compounding growth properties, or when you’re explaining your growth model to investors who need to understand why your unit economics might improve over time rather than staying flat.

If you’re looking for a measurement framework that covers all user lifecycle stages, AARRR gives you more comprehensive coverage. Loops and AARRR aren’t mutually exclusive: AARRR measures what’s happening at each stage, while loop thinking explains the compounding dynamics underneath.

If your goal is to understand overall marketing strategy including brand and demand generation (not just product-led growth mechanics), the Marketing Flywheel might be a better starting point. The flywheel concept, popularized by HubSpot in a marketing context, is essentially a loop built around customer experience and advocacy rather than product-embedded mechanics.

Growth Loops are most valuable as a design lens. When you’re planning a new feature or a new campaign, ask: does this close a loop, or does it just add another step to the funnel? That question alone often changes what you build.

See these frameworks in action: Marketing Case Studies