Growth Loops
A growth loop is a self-reinforcing cycle where each user's behavior produces the input for the next user's acquisition. Here's the formula, the benchmarks, and how to spot one worth amplifying.
Growth Loops
Self-reinforcing cycles where one user's action generates the input that acquires or activates the next user.
A growth loop is a closed system: a user takes an action, that action produces an output (a referral, a piece of content, a review, a shared cart link), and that output feeds back into the top of the system as new acquisition or activation. Unlike a funnel — which is linear and depletes — a loop reinvests every cycle's output into the next cycle's input.
In practice, loops are the strategic frame that sits above individual experiments. A single A/B test on your referral widget is a tactic; deciding that your business runs on a referral loop versus a content loop versus a paid loop is the strategy that tells you which experiments matter.
Funnels assume a fixed pool of traffic at the top: you pay to fill it, then optimise the conversion through each stage. The math is additive — every new customer costs roughly the same to acquire as the last one. That's why CAC keeps drifting upward on most paid channels.
Loops behave differently. Each customer you acquire produces some output — a post-purchase referral, a UGC photo, a review, a shared discount code — that returns traffic or trust to your store without you paying for it again. The interesting question stops being "what's my conversion rate?" and becomes "what's my loop's multiplier, and how do I move it from 0.3 toward 1.0?"
k = (invites_per_user) * (conversion_rate_of_invites)
k
Loop coefficient
The number of new users each existing user generates through one cycle of the loop. k=1 is breakeven (every user replaces themselves); k>1 is exponential growth.
invites_per_user
Outputs per user
How many loop-outputs the average user produces — referrals sent, reviews left, shared links, UGC posts.
conversion_rate_of_invites
Conversion of those outputs
The share of those outputs that turn into a new acquired or activated user.
A Shopify apparel store runs a post-purchase referral widget. The average customer sends 1.8 referral links to friends, and 12% of those links convert into a new order.
Invites per user: 1.8
Conversion rate of invites: 12%
→ k = 1.8 × 0.12 = 0.216
A loop coefficient of 0.216 means the referral loop replaces about 22% of acquired customers organically — meaningful, but not self-sustaining. The store still needs paid traffic to grow; the loop's job is to lower blended CAC, not replace acquisition. The lever to pull next is either invites-per-user (incentive design, prompt timing) or invite conversion rate (landing-page relevance for the friend).
Most online stores don't run on a single loop — they run on a primary loop (usually paid) and one or two secondary loops that subsidise it. The benchmarks below show what loop coefficients look like across the loop types you'll actually encounter at €1M-€15M revenue scale.
Typical loop coefficients (k) by loop type for online retail stores
| Loop type | Typical k | Strong k | Time to cycle | Best fit |
|---|---|---|---|---|
| Post-purchase referral | 0.10 – 0.25 | 0.35 – 0.50 | 2 – 6 weeks | Apparel, beauty, supplements |
| UGC / review loop | 0.05 – 0.15 | 0.20 – 0.30 | 3 – 8 weeks | Visual categories, beauty, home |
| Paid acquisition loop | 0.40 – 0.70 | 0.80 – 1.10 | 30 – 60 days payback | Any store with strong LTV:CAC |
| Content / SEO loop | 0.02 – 0.08 | 0.15 – 0.25 | 6 – 12 months | Considered purchases, gear, electronics |
| Subscription re-order | 0.30 – 0.55 | 0.60 – 0.85 | 30 – 90 days | Consumables, refills, food & bev |
| Influencer / affiliate | 0.08 – 0.20 | 0.25 – 0.40 | 1 – 4 weeks | Fashion, beauty, fitness |
Read these as ranges, not targets. A k of 0.4 on a paid loop only matters if your payback period fits your cash cycle; a k of 0.05 on content compounds for years and may end up your most valuable loop in absolute terms. The right question is which loop your unit economics can afford to feed — and which experiments in your broader experimentation strategy will move that loop's coefficient fastest.
Growth loops: frequently asked questions
A funnel is a one-way path: traffic enters, some fraction converts, and the rest is lost. A loop is closed: the output of one user's journey becomes the input for another's. Funnels deplete and need topping up with paid spend; loops reinvest their own output, which is why they compound.
They're closely related — flywheel is the older marketing term for the same compounding shape. Growth loop is the more precise framing because it forces you to name the specific output (referral, UGC, content, paid LTV) that re-enters the system. A flywheel can be hand-wavy; a loop has a coefficient you can measure.
k ≥ 1 is self-sustaining exponential growth and is rare outside of true viral products. For online retail, a referral loop at k = 0.2-0.3 is solid, and a paid acquisition loop at k = 0.6-0.8 (where LTV reinvested into paid funds the next cohort) is the realistic goal. Anything that meaningfully lowers blended CAC is worth running.
Yes, and most successful brands do. You typically have one primary loop (usually paid, sometimes subscription) and two or three secondary loops (referral, UGC, content) that lower blended CAC. The mistake is treating them as independent — they share creative, audience, and post-purchase real estate, so they compete for attention.
Look at where your last 100 customers came from and trace each one back two steps. If the dominant path is paid ad → first order → repeat, you're on a paid loop. If it's friend's referral → discount code → first order, you're on a referral loop. The dominant path is your primary loop, even if you haven't named it as such.
Loops are the strategic layer that tells you which experiments matter. Within an experimentation strategy, individual A/B tests should be scored by how much they move the coefficient of your primary loop. A 2% checkout uplift on a paid loop with a 60-day payback is worth more than a 10% uplift on a content loop that takes 12 months to cycle.
Usually one of three things: the prompt is timed wrong (asking before delivery, not after a positive moment), the incentive is asymmetric (good for the referrer, weak for the friend), or the friend's landing experience doesn't match what they were promised. Diagnose each step — invites-per-user and conversion-of-invites — separately before changing the offer.
Cycle time varies by loop type. Referral and influencer loops cycle in 1-6 weeks. Subscription re-order loops cycle on their billing cadence (30-90 days). Content and SEO loops cycle in 6-12 months. Don't kill a loop before it's had time to complete two or three cycles — the compounding only shows up after the second one.
They're loops if LTV from existing customers funds the next cohort's acquisition spend without external capital. The output (paid back LTV) re-enters as input (next month's ad budget). They stop being loops the moment you need outside cash to keep them running — at that point they're a funded funnel.
Three things, in order of frequency: a broken handoff in the cycle (post-purchase prompt disappears in a checkout redesign), incentive economics that stop working (rising product cost eats the referral discount), and saturation (your customers run out of friends to invite). Audit each handoff quarterly and watch for k drifting downward cycle over cycle.
Test ideas before you ship them
Run unlimited A/B tests, attach hypotheses to outcomes, and build a searchable archive of what works — and what doesn't.