Funnels spend, loops compound
A funnel is linear: attention goes in the top, some fraction converts, and to get more you put more attention in. Every unit of growth costs another unit of input. A loop closes the circle — the output of the process becomes the input to the next turn — so growth produces the conditions for more growth.
Funnel: spend -> visitors -> signups -> customers (stop spending, growth stops) Loop: new user -> creates an itinerary -> sends it to their client -> client sees it was made with the product -> some clients run agencies too -> new userThree kinds of loop
| Type | Mechanism | Works when |
|---|---|---|
| Viral / word of mouth | Users bring users, directly or as a side effect of use | The product is used with other people |
| Content | Usage or publishing creates pages that get found | There's a real search demand for what you produce |
| Paid | Revenue from customers funds acquiring more | Payback period is short enough to reinvest |
The viral row splits usefully in two. A referral programme is an incentive bolted on — it works modestly and stops when the incentive does. A product where sharing is how the product is used is much stronger, because the loop runs whether or not anyone is thinking about growth.
Writing your loop down
Write it as numbered steps that end where they began. The discipline of closing the circle is what exposes whether you have a loop at all — most first attempts turn out to be funnels with an optimistic arrow drawn at the end.
- Write each step as an action a person takesNot "awareness" or "engagement". "Sends the itinerary to a client" is a step; "builds brand affinity" is not.
- Put a number on every stepHow many people reach it, and what share continue. Estimates are fine; gaps you can't estimate are findings.
- Check that the last step feeds the firstIf it doesn't, you have a funnel. That's allowed — just don't plan as though it compounds.
- Find the lowest-converting stepIn a loop, every step multiplies. The weakest one caps the entire thing.
- Estimate the cycle timeA loop that turns weekly compounds far faster than an identical one that turns quarterly. Speed is a lever people forget entirely.
Improving the right step
Because steps multiply, doubling the worst step does far more than improving an already-strong one. This is the practical payoff of writing the loop down: it tells you where the single week of work should go.
| Step | Now | Improve signup | Improve sharing |
|---|---|---|---|
| Visitors who sign up | 20% | 30% | 20% |
| Signups who create something | 60% | 60% | 60% |
| Creators who share it | 10% | 10% | 20% |
| Recipients who sign up | 5% | 5% | 5% |
| Net loop factor | 0.06% | 0.09% | 0.12% |
Both investments are plausible. Improving the weakest step — sharing, at ten per cent — produces twice the gain of the more obvious signup optimisation, for probably less work.
Common mistakes
- Drawing a loop whose final arrow isn't real, and planning as though it compounds
- Writing steps as abstractions rather than as actions people take
- Optimising the strongest step because it's the most visible
- Bolting on a referral incentive instead of finding the sharing already in the product
- Ignoring cycle time, so a slow loop is treated as equivalent to a fast one
Key takeaways
- A funnel spends input; a loop reinvests output as the next input
- Viral, content, and paid are the three families — the strongest viral loops are how the product is used
- Steps multiply, so the weakest step caps the whole loop
- Cycle time matters as much as conversion, and is more often ignored
Try it yourself
Write your loop as five numbered steps ending where it started, with a percentage on each. If the last step doesn't genuinely feed the first, you've learned that you're running a funnel — which is worth knowing before you plan around compounding.