9. From funnels to loops

§Dave McClure drew the pirate funnel in 2007 and it has organized how startups think about growth ever since: Acquisition, Activation, Retention, Referral, Revenue, the letters spelling the noise a pirate makes. Ward van Gasteren later added Awareness at the top. The funnel is the first thing to learn about growth and, a decade on, the first thing to unlearn.

§ 9.1What each stage of the funnel asks#

§The funnel is useful because each stage is a question, and at a startup the questions are more valuable than the numbers.

§Awareness: how many people do you reach? Acquisition: how many of them show up? Activation: how many take the first important step, the one where they see what the product is for? Retention: how many come back? Revenue: how many pay, and how much? Referral: how many bring someone else?

§The stage that matters first is not the one at the top. It is activation, the first important step, because it is the stage that tests whether the product does the job the discovery said it would. A team that cannot name its activation moment, the specific action after which a user has experienced the value, has not finished discovery. And the stage that matters most is retention, because a product people do not come back to has no business, whatever the top of the funnel shows.

§ 9.2Why a funnel describes a leak#

§Look at the funnel's shape. People enter at the top, and every stage keeps a fraction. Growth in this picture is always bought: more awareness at the top, from marketing spend, content, or luck, to push more people through the same leaky stages. The funnel is linear. Its output does not feed its input.

§Brian Balfour's argument at Reforge, in an essay that changed how growth teams think, is that the companies that compound do not grow this way. They grow in loops: some output of the product, a piece of content, an invitation, a shared artifact, a marketplace listing, becomes the input for the next user. The output of one cohort is the acquisition of the next.

§The difference is not academic. A funnel business needs a growth budget forever. A loop business needs a loop that works, and then needs to make it turn faster. Andrew Chen's The Cold Start Problem is the long version of how loops start: not everywhere at once, but in one small, dense network where the loop can close.

§ 9.3Start with the funnel anyway#

§Given all that, the advice for a startup is still to start with the funnel, for three reasons.

§The funnel's questions come first. Before any loop can close, someone has to be acquired, activated and retained, and the funnel is the checklist for whether those are happening at all. A loop built on a product with no activation moment is a loop that turns once.

§The funnel is honest at small scale. With forty users, a cohort table of activation and retention is readable and true. A loop model with forty users is a story.

§And the loop usually reveals itself. Once retention holds, the team notices where the next users are actually coming from: a shared document, a public page, an invitation, a customer who brought a colleague. That is the loop. It does not have to be designed from the start; it has to be seen when it appears, and then fed.

§ 9.4When to move#

§The signal to shift from funnel thinking to loop thinking is a retention curve that flattens rather than falling to zero: a cohort that keeps a stable fraction of its users after the first weeks. Until then, growth spend fills a bucket with a hole in it. After it, the question changes from "how do we get more people in" to "what does a retained user produce that brings the next one."

§Product Experimentation, the sixth whitepaper, is where the funnel stages and the loop get experiments run against them. This chapter's job was only to make sure the owner knows both shapes before they need to choose. Part III is done. Part IV is about how ownership fails.