Funnels describe how people move through a sequence. Growth loops describe how the output of one cycle creates input for the next.
That difference matters.
A funnel can tell you how efficiently existing demand becomes revenue. A loop can explain how a product, audience, content system or customer base helps generate additional demand over time.
This is why loops are especially useful for thinking about compounding growth.
What is a growth loop?
A growth loop is a closed system in which an action produces an output that can be reinvested to produce more of the original input.
A simplified loop looks like:
input → action → output → reinvested output → more input
For example:
- a user creates useful content;
- the content becomes publicly discoverable;
- a new person finds the content;
- the person signs up;
- the new user creates more content.
The new output feeds the next cycle.
Reforge popularized the distinction between funnels and loops in growth strategy: funnels consume inputs, while loops can generate new inputs through repeated cycles.
That does not mean funnels are obsolete. The two frameworks answer different questions.
Funnels diagnose conversion; loops explain compounding
A funnel is ideal for questions such as:
- Where do visitors drop out?
- What percentage of trials become paid customers?
- Which stage limits revenue?
A loop is better for questions such as:
- How does one cohort create the next?
- Why does growth accelerate or stall over time?
- Which user behavior creates distribution?
- Where does the loop lose energy?
A mature growth system uses both.
The funnel can measure conversion inside the loop. The loop can explain how the system produces its next cohort.
The anatomy of a useful growth loop
A growth loop needs four elements.
1. Input
Something enters the system:
- a new user;
- a customer;
- a piece of content;
- a merchant;
- a listing;
- a partner;
- a dataset.
2. Action
The participant does something that creates value:
- publishes;
- invites;
- buys;
- shares;
- reviews;
- integrates;
- lists inventory.
3. Output
The action produces an asset or signal:
- content;
- a referral;
- a backlink;
- transaction liquidity;
- social proof;
- product data;
- network density.
4. Reinvestment
The output creates or improves the next input.
If the output does not feed back into the system, it is not a true loop.
Common types of growth loops
Viral or referral loops
A user invites another user.
Examples:
- collaboration tools;
- team products;
- referral programs;
- peer-to-peer services.
The core metric is not simply “referrals.” The loop depends on:
- invitation rate;
- acceptance rate;
- activation of invited users;
- time between cycles.
If many users send invitations but few invitees activate, the loop is weak.
Content loops
User or company activity produces content that attracts more users.
Examples:
- public profiles;
- templates;
- reviews;
- marketplace listings;
- programmatic pages;
- editorial content.
A content loop can compound when new content creates new search or social discovery and some of those new users create additional content.
Paid loops
Paid acquisition can also form a loop when acquired customers generate enough contribution margin to finance additional acquisition.
The loop is:
spend → customers → margin → reinvested spend
The important variables are:
- CAC;
- gross margin;
- payback period;
- available capital;
- channel saturation.
If payback takes too long, the loop can be constrained by cash even when unit economics are positive.
Marketplace loops
More supply attracts more demand, and more demand attracts more supply.
Examples include:
- ecommerce marketplaces;
- job platforms;
- service marketplaces;
- ad exchanges.
These loops often depend on local liquidity. A marketplace can appear large globally while still providing a poor experience in a specific geography or category.
Product-data loops
Usage creates data that improves the product, which improves the experience and drives additional usage.
This can occur in:
- recommendation systems;
- personalization;
- fraud detection;
- AI products;
- routing or matching systems.
The loop must create real user value. “More data” by itself is not an outcome.
Measure the loop, not only the endpoint
A loop should have a metric for every transition.
For a referral loop:
- active users;
- invitations per active user;
- invitation acceptance;
- invitee activation;
- time to first invitation from the new cohort.
For a content loop:
- content created per active contributor;
- percentage indexed or distributed;
- qualified discovery per content asset;
- visitor-to-user conversion;
- percentage of new users who create content.
For a paid loop:
- acquisition spend;
- CAC;
- contribution margin;
- payback time;
- reinvestment rate;
- marginal CAC as spend increases.
The key is to identify the weakest edge of the loop.
A loop with five strong transitions and one very weak transition will compound poorly.
Loop speed matters
Two loops can have the same conversion rate but very different growth rates if one cycles faster.
Suppose:
- Loop A completes every 3 days.
- Loop B completes every 30 days.
Even with similar output per cycle, Loop A can reinvest far more frequently.
This makes cycle time a first-class growth metric.
Teams often focus on conversion but overlook latency.
Questions to ask:
- How quickly does a new user create the output that feeds the loop?
- Can onboarding accelerate that action?
- Can approval, indexing or distribution delays be reduced?
- Does the product wait too long before asking for a referral?
Growth loops need quality controls
Compounding low-quality output is not useful.
A content loop can create thin pages that search engines do not value. A referral loop can attract incentivized users with poor retention. A paid loop can scale into low-intent audiences and destroy marginal economics.
The loop therefore needs guardrails.
Examples:
- retained activation;
- contribution margin;
- fraud rate;
- content quality;
- buyer-seller success rate;
- qualified conversion;
- customer satisfaction.
Growth that compounds a negative externality will eventually encounter a constraint.
How to design a growth loop
Step 1: Identify the valuable recurring behavior
Ask:
What do successful users repeatedly do that creates value for themselves and potentially for future users?
This is often more useful than starting with “how can we go viral?”
Step 2: Identify the output
What durable asset, signal or economic value does the behavior produce?
Examples:
- shareable artifact;
- public page;
- review;
- data;
- referral;
- revenue.
Step 3: Connect the output to new input
How does that output reach another potential user or finance another acquisition?
If there is no clear answer, the system may be a funnel rather than a loop.
Step 4: Instrument each transition
Define events and metrics for the loop.
Do not rely on a single top-line growth rate.
Step 5: Measure cycle time
How long does one complete loop take?
Step 6: Add quality guardrails
Define what “healthy growth” means.
Step 7: Test the weakest transition
Improve one edge at a time.
Example: an editorial content loop
A publisher or B2B media company might build:
- research identifies a high-value operational problem;
- the company publishes a strong article or tool;
- search, social, newsletters and partnerships distribute it;
- qualified readers arrive;
- some subscribe or return;
- audience behavior reveals new questions;
- those questions feed the next research cycle.
This is stronger than “publish more articles.”
The feedback from one cycle improves the next.
The key metrics might include:
- percentage of articles receiving qualified discovery;
- subscriber conversion;
- return rate;
- content-assisted revenue;
- percentage of new topics generated from real audience demand;
- time from signal to publication.
Example: a product-led collaboration loop
- a user creates a project;
- the user invites collaborators;
- collaborators receive value;
- collaborators create their own projects;
- they invite additional users.
The loop can fail at multiple points:
- the product asks for an invite before value is clear;
- invitations are ignored;
- invited users cannot activate;
- collaborators remain passive and never create their own project.
“Number of invitations” alone would not diagnose the system.
Loops interact
Most successful products do not have one loop.
They may combine:
- paid acquisition;
- content discovery;
- referral;
- lifecycle retention;
- partner distribution.
These loops can reinforce or compete with one another.
For example, a referral loop may lower blended CAC, allowing more paid acquisition. Paid acquisition may bring more users who create more public content, strengthening organic discovery.
The strategy should identify these interactions.
Do not confuse a loop with a diagram
Growth teams sometimes draw arrows in a circle and call it a loop.
Test the model with three questions:
- Does the output create measurable input for the next cycle?
- Can the loop repeat without an equivalent increase in manual effort?
- Can we measure the conversion and speed of each transition?
If not, the model is probably descriptive rather than operational.
Growth loop checklist
Before investing in a loop, define:
- input;
- value-creating action;
- output;
- reinvestment mechanism;
- conversion at every edge;
- cycle time;
- quality guardrails;
- constraint;
- owner;
- experiment backlog.
Then ask:
- What makes the loop faster?
- What makes it stronger?
- What makes it cheaper?
- What increases output quality?
- What could cause saturation?
Growth loops are not magic. They are systems.
Their advantage is that they force the team to explain where the next cohort comes from and how today’s user behavior changes tomorrow’s acquisition. That makes them a powerful complement to funnels, unit economics and experimentation.
