# How to get users for your AI app: a stage-by-stage playbook

*Show the wow moment, keep CAC in cents, and grow one stage at a time.*

> How to get users for your AI app: what is different about AI apps, the plan by stage, the 3 best early channels, CAC ceiling math and a 30 day checklist.

Source: https://www.appgrowthmarketer.com/playbooks/how-to-get-users-for-my-ai-app
Author: Ar.Bhavesh Panse, AI App Growth Marketer (https://www.arbhaveshpanse.com)
Published: 2026-09-26 · Updated: 2026-09-26

to get users for your AI app, grow it stage by stage: win your first 50 users by hand, prove one short demo video format and one community channel, then add paid spend only behind demos that already work. keep blended CAC low, because every active user costs you inference, and judge every channel on users who come back.

this page is the map. each stage and channel links to a full guide, so read only what matches where you are. the deeper reasons AI apps behave differently are in [AI app growth marketing](https://www.appgrowthmarketer.com/learn/ai-app-growth-marketing).

## What is different about getting users for an AI app

four things change the plan for an AI app. every active user costs you inference, so free users are not free. people try an AI app once and leave if the first result is weak. the product has a wow moment you can show on video in seconds. and copycats can ship your core feature within weeks.

| what is different | what it changes in your plan |
|---|---|
| inference cost per user | judge channels on returning or paying users, never installs |
| try and leave behaviour | fix the first session before you buy any traffic |
| a demo-able wow moment | short demo video is your strongest first channel |
| fast copycats | build an audience you own early, because features get copied and audiences do not |

the try and leave problem shows up in the data. RevenueCat's 2026 State of Subscription Apps report, [as covered by TechCrunch](https://techcrunch.com/2026/03/10/ai-powered-apps-struggle-with-long-term-retention-new-report-shows), found AI app subscribers cancel about 30% faster than non-AI apps at the median.

## The plan by stage

the plan changes at each stage because the job changes. under 100 users you are learning, so you work by hand. from 100 to 10,000 you are finding one channel that repeats and a creative format that wins. past that you are scaling what works without letting CAC or inference cost run away. skipping a stage usually means wasted spend.

| stage | your main job | AI app twist | full guide |
|---|---|---|---|
| 0 to 50 users | find the aha moment by onboarding people yourself | time the seconds from open to first AI result, and cut them | [first 50 app users](https://www.appgrowthmarketer.com/playbooks/first-50-app-users) |
| 50 to 100 | work two channels every day | ask new users what they expected to get | [first 100 app users](https://www.appgrowthmarketer.com/playbooks/first-100-app-users) |
| 100 to 1,000 | find one channel that repeats | set free usage limits before any launch or paid test | [first 1,000 app users](https://www.appgrowthmarketer.com/playbooks/first-1000-app-users) |
| 1,000 to 10,000 | build a weekly creative engine, test small paid | track inference cost per active user next to CAC every week | [first 10,000 app users](https://www.appgrowthmarketer.com/playbooks/first-10000-app-users) |
| 50K to 1M | scale proven channels, add markets | protect margins as heavy free users multiply | [50K to 1 million app users](https://www.appgrowthmarketer.com/playbooks/50k-to-1-million-app-users) |

one proof that the order matters: ZuAI, an AI study app, went from 10K to 2M users in eleven months at a $0.02 blended CAC, through TikTok UGC, Reddit and paid acquisition. raw phone-shot videos beat studio ads, and paid spend only scaled behind creatives that had already proven themselves. details: [ZuAI case study](https://www.arbhaveshpanse.com/case-studies/zuai).

## The 3 channels that fit AI apps best early

short demo video, Reddit and niche communities, and launch sites fit AI apps best in the early stages. demo video shows the wow moment in seconds. communities put you where people already ask which AI tool to use. launch sites send a burst of curious early adopters who enjoy trying new AI products. all three cost time rather than money.

### 1. Short demo video

a demo video shows a real problem, one tap in the app, and the finished result, in under 20 seconds. film or screen record it on a phone. post one a day on TikTok and Reels for at least three weeks before judging, because a few posts tell you nothing. cost: about an hour a day, no money.

> "hook: the problem on screen in the first 2 seconds. action: one tap in the app. result: the real output, full screen, no mockups. caption: what it did, in the words your users use. length: under 20 seconds."

when one video wins, TikTok's [Spark Ads](https://ads.tiktok.com/help/article/spark-ads) let you put paid spend behind that same organic post. the full daily routine is in [first 100 users from TikTok](https://www.appgrowthmarketer.com/playbooks/first-100-users-from-tiktok).

### 2. Reddit and niche communities

people ask "which AI tool does X?" in subreddits, Discord servers and forums every day. answer the question fully first, say plainly that you built the app, and follow each community's own rules on self promotion. cost: about an hour a day. step by step guides: [first 100 users from Reddit](https://www.appgrowthmarketer.com/playbooks/first-100-users-from-reddit) and [first 100 users from communities](https://www.appgrowthmarketer.com/playbooks/first-100-users-from-communities).

> "i had the same problem with [task]. what worked for me: [two useful steps anyone can do without an app]. full disclosure, i built [app] for this, so i'm biased, but it's free to try if the manual way gets old."

never use fake accounts, vote rings or copy pasted replies. they get accounts banned.

### 3. Launch sites

Product Hunt and Hacker News can send a burst of curious visitors in a single day. read the rules first. Product Hunt's [launch guide](https://www.producthunt.com/launch) says you cannot ask people directly for upvotes. Hacker News [Show HN rules](https://news.ycombinator.com/showhn.html) say people must be able to try it, ideally without signups, and ban landing pages. full plan: [first 100 users from launch sites](https://www.appgrowthmarketer.com/playbooks/first-100-users-from-launch-sites).

AI warning: set a daily free usage cap before launch day, or a good launch becomes your biggest inference bill.

## CAC ceiling math for AI apps

your CAC ceiling is the most you can pay to acquire a user and still break even once inference is counted. in my experience consumer AI apps struggle above roughly $0.50 blended CAC, and install costs of $3 to $15 kill an app whose early users are worth cents. work out your own ceiling before you spend anything on ads.

blended CAC is all acquisition spend divided by all new users, across every channel. free channels pull it down and give paid room to test.

1. take average revenue per new user over 90 days, including users who never pay
2. subtract average inference cost per new user over the same 90 days
3. what is left is your break even CAC. set your written ceiling well under it
4. check blended CAC against it weekly, before raising any budget

illustrative only, not from a real account:

| channel | spend | new users | cost per user |
|---|---|---|---|
| TikTok demo videos (organic) | $0 | 900 | $0 |
| Reddit and communities | $0 | 300 | $0 |
| paid behind a proven demo | $400 | 800 | $0.50 |
| blended, all channels | $400 | 2,000 | $0.20 |

paid alone sits at the edge. blended with organic, it has room. the fixes is in [how to lower CAC for a consumer AI app](https://www.appgrowthmarketer.com/paid/lower-cac-consumer-ai-app).

## Your first 30 days checklist

the first 30 days are about one working first session, one demo format that people watch, and one community where you are already known as helpful. it takes about two hours a day and very little money. do the weeks in order, because traffic from week three is wasted if the week one fixes are not done.

**week 1: fix the product side**
- [ ] cut every screen between install and the first AI result that is not strictly needed
- [ ] add a ready made example so new users never face a blank prompt box
- [ ] set free usage limits and a daily cap on the most expensive feature
- [ ] write down your CAC ceiling using the four steps above

**week 2: start two channels**
- [ ] post one demo video a day
- [ ] write useful replies in two or three communities, disclosing that you built the app
- [ ] message every new user within a day

**week 3: double down**
- [ ] turn your best demo into three new hooks
- [ ] spend more time in the community whose users came back
- [ ] prepare a launch site post with a short demo and a working free tier

**week 4: launch and decide**
- [ ] launch on one site and answer every comment that day
- [ ] keep the best channel daily, then test a small capped paid budget behind your best demo

## What to measure

measure returning users by source, seconds to the first AI result, inference cost per active user, and blended CAC against your ceiling. at this stage, installs and views mislead you, because a viral demo can bring thousands of tourists who never come back. the number that matters is users who return, per hour or per dollar spent.

| metric | how to track it | why it matters |
|---|---|---|
| day 7 return rate by source | analytics plus a "how did you find us?" question | shows which channel brings real users |
| seconds to first AI result | time a fresh install weekly | the first session decides who stays |
| inference cost per active user | model provider bill divided by weekly active users | tells you what a free user really costs |
| blended CAC vs ceiling | total spend divided by total new users | the line you must stay under before scaling |

when one channel sends users who come back and blended CAC stays under your ceiling, move to the next stage guide.

## Frequently asked questions

### How do I get users for my AI app?

grow it in stages. win your first 50 users by hand, then pick two channels that fit AI apps, usually short demo video plus Reddit or niche communities, and work them daily. add a launch site as a one day push. only put paid spend behind demos that already work organically, and judge everything on users who come back.

### How much should I pay to acquire a user for a consumer AI app?

less than you think. every active user costs you inference, so your ceiling is lower than for a normal app. in my experience consumer AI apps struggle above roughly $0.50 blended CAC. work out what a user earns you in 90 days, subtract what they cost to serve, and stay well under what is left.

### What is the best way to market an AI app with no budget?

film short screen recordings of the app turning a real problem into a finished result, and post one a day on TikTok and Reels. at the same time, answer questions in the subreddits and communities where your users ask for help, saying openly that you built the app. both cost time, not money.