Growth Loop Calculator

A free tool for founders who want to know whether a referral, content or paid loop actually compounds. You enter the stages of one turn of the loop with your own conversion rates, and what one completed turn produces, such as the invites each user sends. It returns the loop factor k, how far the loop stretches every user you acquire, your effective CAC once the loop is counted, and the stage holding the loop back. With starting users and a paid feed, it projects users turn by turn against paid acquisition alone.

What it computes

The loop factor k. Every stage rate multiplied together, times the output per turn. For a referral loop this is the viral coefficient: three invites per user, with 20% of invites becoming a user, gives k = 0.6.
Amplification. Below k = 1, every user you acquire ends up as 1 / (1 - k) users, because each user brings k more, and they bring k more again.
Effective CAC. Paid CAC times (1 - k): the same spend, spread across every user the loop adds.
The throttle. The stage with the lowest rate. Because k is a product, the same 10-point gain raises k most at the lowest rate. The tool shows what that gain does at every stage.
A projection. Each turn, the users who arrived last turn bring k times their number and the paid feed adds a fixed number on top, set against starting users plus the paid feed alone. Tell it how many days a turn takes and the projection is labelled in days.

The verdict

Viral at k of 1 or more: the loop sustains itself until the market fills. Strong sub-viral from 0.5 to below 1: the loop more than doubles what each paid user is worth. Contributing from 0.2 to below 0.5: a real assist that lowers effective CAC but will not carry growth alone. Weak below 0.2, and broken at 0, when nothing feeds back.

Why I built it around k, not the funnel

A funnel asks how many people reach the bottom. A loop asks whether one turn produces enough to start the next one, and you can't see that if you model referrals as one more line in the funnel. A k well below 1 is easy to dismiss as "not viral". I'd look at it the other way: a loop at k = 0.4 makes your real cost per customer 40% lower than the paid number says. So I'd rather a founder fix the weakest stage of a sub-viral loop and feed it with paid acquisition than chase k of 1 or more.

Limits

Optimism compounds. A rate padded at each stage multiplies into a k the loop will not deliver, so use measured rates and treat estimates as a scenario.
The projection holds k constant and sets no saturation ceiling. Real loops tend to slow as the network fills, and at k of 1 or more the curve runs away.
One period is one full turn of the loop. The tool does not know how long a turn takes unless you tell it.
It has to be a loop: the last stage's output must start the first stage again. If nothing feeds back, k means nothing.

The same engine runs as an agent skill, aaj-marketing-loops. To design the referral offer itself, read Referral Programs for Startups. For loops built into the product, see the Product-Led Growth Playbook. To see what effective CAC does to payback, use the Unit Economics Calculator.

Run this in your agent

The free AAJ agent skill Marketing Loops does this job inside a coding agent, on your own data. Install it with npx skills add sarojkjha/aaj-marketing-skills --skill aaj-marketing-loops.

Using Claude or ChatGPT instead? Add the AAJ engines as a custom connector, with no install: https://aaj-engines-mcp.sarojkjha10.workers.dev/mcp