A funnel needs constant refilling. A loop reinvests its own output. The difference decides whether growth compounds.
12 min read · Level 3
A funnel needs refilling from outside for ever; a loop reinvests its own output. But loops sit on top of retention, and a loop built over a leaking bucket just brings more people in to leave — amplifying churn rather than growth. Chasing virality in a product with no natural sharing moment produces an invite prompt everyone dismisses and a coefficient near zero.
A funnel takes people in at the top and loses them at each step — it needs refilling from outside for ever.
A growth loop feeds its own input: using the product produces something that brings in new users.
Retention is the foundation of both. Without it, everything you acquire drains away.
Because retention determines whether growth compounds or merely accumulates. At 5% monthly churn you replace your whole customer base in under two years — running hard to stand still.
And because loops are the difference between growth that needs a bigger budget every month and growth that gets cheaper over time.
A funnel business: spend $5,000 on ads → 25 customers → 5% monthly churn. To grow, spend more every month. Stop spending and growth stops immediately.
A loop business: users publish public rota templates → those pages rank in search → other ops managers find them → some sign up → some publish templates.
The loop's output is its own input. It gets stronger with scale, and it does not stop when you stop paying.
Three common loops:
Most B2B companies will not have a strong viral loop and should not pretend otherwise. A referral loop with a k of 0.4 — every ten customers bringing four more — is entirely achievable and cuts effective CAC by 40%.
First-time founders often chase virality for products that have no natural sharing moment. Forcing an invite prompt into a single-player tool annoys users and produces a k near zero.
The second mistake: building loops before fixing retention. A loop on top of a leaking bucket amplifies churn — you bring in more people who also leave.
The third: not asking for referrals because it feels like begging. Asking a happy customer for two introductions, at the moment they got value, is normal and effective.
Until the curve flattens, everything else is filling a leaking bucket faster.
Is there a point where using the product involves someone else? That is where a loop can exist. If there is not, do not force one.
A public template, a shared report, a profile. Something with standalone value for someone who has never heard of you.
At the moment of value, not at renewal. Two named introductions, not 'let me know if you know anyone'.
New users generated per existing user. Even 0.3 to 0.4 is valuable — it reduces effective CAC substantially.
Above 100% means the company grows with zero new customers. It is the strongest loop of all and the least discussed.
After product-market fit, once retention is stable enough to build on.
Do not build loop mechanics pre-PMF. And do not add referral incentives while churn is high — you will pay to acquire people who leave.
Dropbox's referral programme gave extra storage to both the referrer and the person invited. It is widely cited because the incentive was the product itself, cost the company almost nothing marginally, and made both parties more invested. Bolting a cash incentive onto a product with no such lever rarely reproduces the result.
The best referral incentives are more of the product. If yours has no such currency, a referral loop will be far weaker — plan accordingly rather than assuming.
Founders dismiss viral coefficients below one as failure. In practice a steady 0.4 — every ten customers bringing four more — reduces effective acquisition cost by around forty percent permanently, which is a larger and more durable gain than most channel optimisation produces.
Do not aim for viral. Aim for measurable referral, ask for it systematically at the moment of value, and count it.
Create a free account to read the rest and apply it to your own startup.