What 58 ad signups did in their first days
Two Instagram video ads, three dollars a day each, five days. 58 signups, 48 projects, 22 working apps, zero payments. All the numbers, including the ugly ones.
Ilyos Olimov · · 4 min read
Until October 15 AppX got one to four signups a day, mostly from search and directories. I wanted to know what happens when strangers arrive from an ad instead. So I ran the smallest test I could.
The setup
Two vertical videos, one ad set each, Instagram Stories and Reels only, three dollars a day per set, October 15 to 20.
- "Goya", in Uzbek, targeted at Uzbekistan.
- "Idea", in English, targeted worldwide. On day three I narrowed it and excluded India and Pakistan.
The goal was landing-page views, not signups. I did not trust the pixel enough to optimise for registrations yet.
Meta rejected the English ad twice. The first time for a "deceptive link" after I attached pixel tracking, which was lifted on its own the next morning. The second time after I changed the audience, and that one stuck. So the English ad ran for roughly two and a half days, the Uzbek one for all five.
What the ads did
By the evening of October 18:
| Uzbek ad | English ad | |
|---|---|---|
| Impressions | 11,028 | 10,652 |
| Link clicks | 201 | 130 |
| Landing page views | 142 | 95 |
| Spend | $9.75 | $4.90 |
| Registrations Meta attributes | 10 | 3 |
Total spend under twenty dollars for the whole test. Click rate about 1.8 percent on the Uzbek ad, around 1.2 on the English one.
What the signups did
This is the part no ad dashboard shows you. From October 15 to 19 the database recorded 58 new accounts, against the usual one to four a day.
| Count | |
|---|---|
| Signed up | 58 |
| Signed up with Google | 52 |
| Created at least one project | 34 |
| Got at least one working app | 22 |
| Spent credits | 23, on average 47 of their 100 |
| Hit at least one failed build | 14 |
| Paid | 0 |
48 projects in total. Mostly Uzbek apps: a clothing size helper, a personal finance tracker, a farm ledger, a milk delivery shop, a school test app, a manga reader.
The three things I got wrong
The paywall was on for the first day and a half. Before the test, AppX asked new users to pay five dollars before their first build. The first 36 ad signups hit that wall and left. I turned it off on October 16 and gave every one of them 100 credits. Of those 36, one came back.
Edits were broken for a week and I did not know. Fourteen of the 58 had a failed build. Ten of those failures were the same bug: a validation step parsed JSON files as TypeScript and rejected every edit that touched app.json. Nobody writes to you about this. They just stop. I found it in the failure codes on day three and shipped the fix on day four.
A deploy refused builds for an hour. On October 19 my own deploy drained for 68 minutes and one user tried three times to build during exactly that window. That story has its own post.
Was it worth it?
Under twenty dollars bought 58 signups, 22 working apps and a list of three real product bugs that a hundred quiet days would not have surfaced. Nobody paid. For a product at this stage I think that is the right outcome: the money was not the point, the behaviour was.
Two days after the ads stopped, signups went back to zero. That is the clearest number in this whole post. Content and directories bring a trickle. Right now the ads are the only tap.
Next
Run the Uzbek ad again, longer, with the bugs fixed and the paywall off. Optimise for registrations this time, since the pixel now sees them. And keep reading the failure codes every morning.
Keep going
More practical ways to move your app forward.
- How to describe your app to AI: a prompt template
Write a clear AI app-builder prompt with a copyable template, a practical example, acceptance checks, and focused follow-up edits.
- How to turn a client brief into a mobile app prototype
Turn a client brief into a mobile app prototype with a copyable AppX prompt, a practical test checklist, and a focused client-feedback workflow.
- JSON is not TypeScript: the one-line bug that failed ten edits
For a week, every edit that touched app.json failed with a syntax error. The JSON was valid. The validator was parsing it as TypeScript. How I found it from failure codes, and the second bug hiding behind it.
- An Uzbek AI assistant prototype in ten minutes
A user's prompt, one plan, 26 credits, ten minutes. What came out, what is real, what is honestly a placeholder, and why the prompt worked.
- Do not pretend the feature works
A user wrote one line in his prompt that every prototype should carry: do not pretend that AI voice, memory or autonomous actions work unless they are actually implemented. Why an honest prototype is worth more than an impressive one.
Ready to try your idea?
Start with a clear description. AppX will help you turn it into an app you can test.