A funnel describes a linear sequence a user moves through once: awareness, acquisition, activation, conversion. It's a genuinely useful tool for diagnosing where a single pass of users drops off. But a funnel has a built-in assumption that's easy to miss — it starts from traffic that arrives from somewhere outside the system, usually paid or organic acquisition the team has to keep refilling from scratch every period. A funnel can be optimized to convert a higher percentage of the traffic poured into the top, but it has no mechanism, by itself, for generating the next period's traffic. That's the structural ceiling: funnel optimization is additive, not compounding — a 10% improvement this quarter doesn't make next quarter's traffic any cheaper to acquire.

What makes something a loop instead

A growth loop is a closed system where the output of one cycle becomes the input to the next cycle, without requiring a new round of external spend to restart it. The canonical structure has four stages: an input (some unit of resource — a user, a piece of content, an invite), an action the input performs, an output that action generates, and a reinvestment step that feeds that output back in as a new input, ideally at a larger scale than before. Because the output of one pass becomes the input to the next, growth from a working loop compounds — each cycle's output is partly determined by the accumulated results of every prior cycle, not just that period's fresh spend.

growth_loop_structure.txt

text

          +------------------+
          |      INPUT       |  <-- existing user / piece of content
          +------------------+
                   |
                   v
          +------------------+
          |      ACTION      |  <-- user invites a friend / shares content
          +------------------+
                   |
                   v
          +------------------+
          |      OUTPUT      |  <-- new signup / new pageview from share
          +------------------+
                   |
                   v
          +------------------+
          |   REINVESTMENT   |  <-- new user becomes a NEW input,
          +------------------+      repeats the cycle at larger scale
                   |
                   +------------------> back to INPUT (loop closes)

Three loop archetypes worth knowing

Most real growth loops reduce to a small number of recurring patterns. A viral/social loop turns existing users into the acquisition channel for new ones — Dropbox's referral storage bonus, a collaboration tool where inviting a teammate is required to use the product at all. A content loop turns user activity into indexable content that pulls in new organic search traffic — Pinterest pins, Yelp reviews, and Stack Overflow answers are all generated by existing users but discovered by new ones via Google. A paid loop reinvests revenue directly into acquiring the next cohort — an e-commerce business that pours a fixed percentage of each cohort's margin back into ads, where a higher-LTV cohort funds a bigger next spend without any new external capital. Note that a paid loop still needs a funnel inside it (the conversion mechanics of the ad click to purchase) — loops and funnels aren't opposites, a funnel is often the engine running inside one stage of a loop.

Funnels vs. growth loops

Dimension	Funnel	Growth loop
Shape	Linear, one-directional	Circular — output feeds back as input
Growth pattern	Additive, resets each period	Compounding across cycles
Primary question	What % converts at each step?	Does each cycle produce a bigger input than the last?
Dependency	Needs continuous external traffic	Self-sustaining once the loop closes, needs less new spend
Failure mode	Drop-off at a specific step	Loop coefficient below 1.0 — it shrinks instead of compounds

A loop coefficient under 1.0 is just a slow-decaying funnel

Not every loop-shaped mechanism actually compounds. If, on average, each cycle produces less than one new unit of input for the next cycle (loop coefficient < 1.0), the loop decays rather than grows — it's structurally a funnel with extra steps, not a compounding engine. The entire point of measuring a loop is checking whether that coefficient clears 1.0, ideally with real margin above it, since real-world loops also face fatigue (the same message loses effectiveness as a network saturates).

Loops don't replace funnels — they contain them

Teams sometimes treat "growth loops" as a replacement for funnel thinking entirely, which overcorrects. A viral loop's action step (inviting a friend) still has conversion mechanics worth funnel-optimizing — what fraction of users who see the invite prompt actually send one. Loop thinking tells you where to look for compounding leverage; funnel thinking is still the right tool once you're inside a single stage of the loop.

Diagnosing which one you actually have

A quick test: stop all new paid/organic acquisition spend for one period and see whether growth continues. If it drops to roughly zero, the business is running on a funnel fed by external traffic, full stop — there's no loop, no matter what the pitch deck calls it. If growth continues (even at a reduced rate) because existing users keep generating new ones through sharing, content, or reinvested revenue, a real loop is operating underneath. Most mature growth programs run a portfolio of several loops plus funnel-optimization work layered on each — the point isn't to pick one model over the other, it's to know which compounding mechanisms actually exist today versus which metrics are just funnel efficiency dressed up in loop language.

What's next

Diagnosing whether a loop is actually compounding requires picking the right single metric to track its health against — that's the role of a North Star Metric.

Next: The North Star Metric Framework →

I build these systems professionally.

Whether it's a RAG pipeline, analytics migration, or AI workflow — let's talk.

Work With Me →

Need production AI or MarTech expertise? I build custom LLM workflows, server-side GTM tagging, and analytics infrastructures professionally.

Need custom AI or MarTech setup? Let's build together.

Let's talk →