Growth Loops: Build Acquisition That Compounds

Why this matters

The funnel is the most useful diagram in marketing, and also the most misleading one for planning growth. It shows traffic entering at the top and customers leaving at the bottom. What it hides is the arrow back. In a funnel, the customer is the end of the story. In a loop, the customer is the start of the next cycle.

That difference decides how your growth behaves over time. A pure funnel business grows in line with spend. Stop the ads, stop the outbound team, stop the launch campaigns, and new customers fall to near zero. A loop business keeps producing new users from its existing users, its content or its revenue. Each cycle hands fuel to the next.

You can see this in public numbers. Zoom’s S-1 says most large customers started with a free host who had been invited to a meeting. Dropbox’s founder Drew Houston told the 2010 Startup Lessons Learned conference that paid search cost them 233 to 388 dollars per acquisition for a 99 dollar product, while a referral program permanently lifted signups by 60%. Neither company out-spent its way to scale. They built loops.

This lesson gives you a working model of loops, the four main types, and a method to map and quantify your own.

The framework

Funnels vs loops

Funnels are not wrong. Every loop contains a funnel. The loop view simply adds the return arrow and forces you to measure it.

Anatomy of a loop

Every loop has four parts. If you cannot name all four, you do not have a loop yet.

1. Input. A new user, a new piece of content, or a unit of money.
2. Action. What that input does inside your product. A user invites a teammate. A seller lists a product. A customer places an order.
3. Output. The result that has value outside the product. An invite email. A public page that search engines index. Gross profit.
4. Reinvestment. How the output becomes the next input. The invitee signs up. A searcher lands on the page and registers. Profit is spent on the next ad.

Two numbers describe any loop: gain (how many new inputs each input creates) and cycle time (how long one turn takes). A loop with gain 0.5 and a 7-day cycle can beat a loop with gain 0.8 and a 120-day cycle inside a single quarter.

The four loop types

Here is the viral loop in its simplest form, as Zoom runs it.

And here is a paid loop, where money is the thing that cycles.

Step-by-step playbook

1. Start from natural behaviour, not from a tactic list. Write down what your best users already do that touches non-users. Do they send invoices? Share dashboards? Post photos of the product? Leave reviews? Your strongest loop is usually an amplified version of something people already do. Forcing a share step people never asked for produces low gain.

2. Write the loop as a sentence with four blanks. “Each [input] does [action], which produces [output], which brings in [next input] through [reinvestment].” If any blank needs the word “hopefully”, go back to step 1.

3. Instrument every step with a cohort and a window. For each step, define the numerator, the denominator and a time window. Example: “Share of accounts created in week W that send at least one invite within 14 days.” Tag loop traffic at the source: invite tokens, referral codes, UTM parameters on shared links, a page-type label on UGC pages.

4. Calculate gain and cycle time from real data. Multiply the step rates for a recent cohort. Measure cycle time as the median days from input to next input. Do this for at least three cohorts to see the trend.

5. Find the weakest step and fix that first. Because the steps multiply, a 50% lift at any step lifts the whole loop by 50%. Go for the step with the most headroom and the cheapest fix. Usually that is the step where users decide whether to take the action at all.

6. Shorten the cycle. Move the sharing moment earlier. Get pages indexed faster. Pick a product or offer with a faster second purchase. A shorter cycle means more turns per year.

7. Choose one dominant loop. Most companies that scale with loops have one main loop and one or two supporting loops. Zoom pairs a viral loop with a sales loop that converts viral accounts into enterprise contracts. Run the second loop only after the first is instrumented and improving.

8. Review the loop monthly with finance. Gain, cycle time and the share of new users that come from loops belong in the same monthly review as CAC and payback. Loops change your blended CAC, so they are a finance topic too.

Worked example

Say a B2B invoicing tool for freelancers adds “Sent with InvoiceApp, create yours free” to every invoice. Here is one month’s cohort.

Loop gain g = 0.60 × 8 × 0.05 × 0.20 = 0.048.

So every 1,000 accounts bring in 48 more, which bring in about 2 more. Total from the seed is 1,000 ÷ (1 − 0.048), about 1,050. Real, but small.

Now compare two fixes.

• Fix A: a better footer. Click-through goes from 5% to 7.5%. New gain = 0.60 × 8 × 0.075 × 0.20 = 0.072.
• Fix B: a guided first invoice in onboarding. Senders go from 60% to 85%. New gain = 0.85 × 8 × 0.05 × 0.20 = 0.068.

Both lift the loop by roughly 40 to 50%. Doing both gives 0.85 × 8 × 0.075 × 0.20 = 0.102, more than double the original. That is the core idea: improvements multiply.

For a D2C paid loop, swap users for money. Say CAC is ₹800 and the average new customer produces ₹560 of gross profit in the first 60 days. Gain = 560 ÷ 800 = 0.7. Each cycle returns 70 paise per rupee. Add a replenishment reminder that raises 60-day gross profit to ₹880, and gain becomes 1.1. Now the same budget grows itself every 60 days, as long as CAC holds when you scale.

B2B SaaS vs D2C: what changes

How to measure it

• Loop gain. Product of step rates for a defined cohort. Report it per cohort, not as a running average, so you can see the trend.
• Cycle time. Median days from an input entering to the next input it creates (for example signup of inviter to signup of invitee).
• Loop-sourced share. New users or customers tagged as loop-sourced ÷ all new users or customers, per month. Watch it alongside total volume.
• Step rates. Each with numerator, denominator and window. Example: invites accepted within 14 days ÷ invites sent.
• Segment. Split by acquisition source, plan, country and device. Loops often run hot in one segment and cold in others.
• Blended CAC. Total acquisition spend ÷ all new customers. A working loop lowers it even when paid CAC stays flat.

Templates

Copy this loop map for each loop you run: