AI app user acquisition: the channel mix that took ZuAI from 10K to 2M

AI app user acquisition has one rule the generic guides skip: organic is not a channel, it is the testing ground for every channel you will ever pay for. nothing at ZuAI got budget until it had already earned attention for free. that single rule is why the blended CAC ended at $0.02 and not $2.

here is the mix, in the order it was actually built, with the numbers at each step. the short version of who i am is on the AI app growth marketer page. this post is the how.

Step one: the problem nobody tells you about

ZuAI was an AI study mentor for students. the users were under 18. and you cannot target anyone under 18 on any ad platform. no interest targeting, no lookalikes, nothing.

so the only door open was organic, and i had no idea which door it was. that constraint turned out to be the best thing that happened to the account, because it forced the method that later scaled.

Step two: volume finds the hook, not taste

i shot 30 videos in one month to find out what worked. i could not afford creators, so a college junior recorded the first ones on his phone.

two or three moved. the rest died. the two that moved had one thing in common: the thumbnail said “under 18”. that was the entire secret. kids under 18 want content made for people under 18. one video did 2.1 million views and 120,000 users. the next did a million views and 40,000 users.

that is what “user acquisition” means at this stage: 30 attempts, two hooks, and the discipline to notice which two. no brief would have written “under 18” on a thumbnail. volume found it.

Step three: unlisted winners become the ads

here is the part most AI app user acquisition guides miss entirely. we uploaded the winning videos as unlisted on youtube and ran them as ads, so only the target audience ever saw them as ads. the public channels stayed clean: free, funny, genuinely useful content. the acquisition machine ran underneath, invisible.

the video shot by a broke college junior three years ago is still the best performing ad in that account. it never had to look like an ad because it never was one first.

Step four: the channel that broke, and the one that did not

we took the exact same creative to meta. meta gave us installs, lots of them, and almost none of them paid. we killed it.

tiktok, with a different structure: five separate accounts, one creator each, $1,000 per creator for 30 videos in 30 days. the first creator got views and nothing else and was replaced. the second posted near identical videos, roughly a thousand views each, week after week. then one did 2 million.

the difference, frame by frame: she was not showing the app. she was doing a normal thing in a normal day, talking to the camera while eating fruit, and the app happened to be there. the moment we stopped advertising, it started working.

we downloaded that video, ran it as an ad at $20 a day, then $50, then $100, then $200, and scaled the channel to $10,000 a month.

Step five: the expensive mistake

around the same time we decided to do it properly. the top creator in the education space, $2,000 for one video. it got 12,000 views and close to zero users.

the most expensive asset in the entire account failed while a phone video from a junior did millions. that is when i stopped believing in production value and started believing in volume. ugly ads beat polished ads, and i have three years of spend since saying so.

Step six: reddit for the first 10,000, not the last

everyone knows their users are on reddit. everyone gets destroyed trying. i got 40+ accounts banned learning what reddit actually enforces, as opposed to what the sidebar says it enforces.

the lesson: posting and commenting reads as promotion no matter how carefully you write it. direct messages do not. reply rate: 30 to 35 percent, the highest of any channel i have ever run.

it does not scale and it does not need to. reddit is how you get your first 10,000 users and, more importantly, the exact words your customers use to describe their own problem. paid tells you what converts. reddit tells you why anyone cares. the method is public as the SAFE Reddit playbook.

The CAC maths, honestly

11 months later: 2 million users, $0.02 blended CAC, 35,000 daily active users, 10 percent converting to paid.

that $0.02 is blended, and the word matters. organic brought most of the users at zero cost; paid scaled proven creative at a real cost; the blend landed at two cents. anyone quoting that number as a paid CAC is misreading it, and anyone promising it for your app before seeing your retention is guessing.

the honest sequence for any AI app: measure retention, run organic as the lab, promote the winners to paid, use reddit for the first users and the language, then scale what already works. the growth playbooks hold each step in detail, free, no email form in front of them.