11. Acquisition experiments

§The question: which single channel brings users who activate and retain, at a cost the economics accept?

§Acquisition is where growth spending happens, so it is where the discipline matters most. The validation signals gave you permission to spend; this chapter is about spending it on finding a channel rather than scattering it across all of them.

§ 11.1One channel before many#

§Gabriel Weinberg and Justin Mares catalogue nineteen channels, and their central point is that a startup succeeds through one of them, not all. Their bullseye framework is the method: brainstorm every channel, pick the two or three most promising, run cheap tests on those, and then commit to the one that works. Presence everywhere is the instinct and the mistake; depth in one is what grows a company.

§The reason is focus and the reason is learning. Nineteen shallow channel efforts produce nineteen ambiguous results. Three real tests produce a clear winner, and the winner rewards depth: a channel that works at small spend usually works at larger spend, and the compounding comes from mastering one channel's mechanics, not from adding a tenth.

§ 11.2Measure activated, retained users#

§The metric for an acquisition experiment is not cost per sign-up. It is cost per activated user who retains, because a channel that delivers cheap sign-ups who never activate is delivering nothing, and the funnel from the previous chapter is why: a sign-up that leaks out at activation cost money and produced no user.

§This changes which channels win. A channel with expensive clicks but users who activate and stay can beat a channel with cheap clicks whose users vanish. You cannot see this in the acquisition number alone; you have to follow each channel's cohort down the funnel, which means the instrumentation from chapter 17 has to tag users by channel from the start.

§ 11.3The anatomy applies#

§Each channel test is an experiment with the Part I anatomy. Hypothesis: this channel brings users who activate at this rate for this cost. Metric: cost per activated user from the channel. Threshold: the cost the unit economics permit, which comes from the fifth whitepaper's lifetime value. Decision rule: below the threshold, go deeper; above, drop the channel and test the next.

§Andrew Chen's law of shitty clickthroughs is the caution that makes duration matter: every channel decays as it saturates and as users habituate, so a channel that works today works less tomorrow, and an acquisition strategy is a pipeline of channel experiments, not a single win to ride forever.

§ 11.4Channels and the loop#

§The best acquisition channels are the ones that close a loop. Paid acquisition is a funnel: it works while you pay. A channel where acquired users produce the next users, content they publish, invitations they send, public artifacts they create, is a loop, and it is worth more than a paid channel at the same cost because it compounds. When testing channels, weight the ones that could become loops, because the goal from the previous chapter is an engine, not a pump.

§ 11.5What you leave with#

§Two or three channels tested properly, one at a time, measured by cost per activated and retained user against the economics threshold. One channel identified for depth, weighted toward channels that could close a loop. And every user tagged by channel from arrival, so the funnel can be read per channel. The next chapters move down the funnel to activation, retention and referral.