Note on 2 weeks. Vibe Corder Diary 2(How to make APP marketing test with Auto Video pipeline)

Follow-up to the earlier post. Two-week update. what shipped, what the numbers were, what the caveats are.

(Written in Korean, translated paragraph by paragraph with Google Translate.)

The Number

Six apps are live across iOS and Android. A seventh is close to release. The video pipeline runs on its own, so app development has not been slowed by it.

The landing page went up recently. It has outbound tracking links attached. The download counts shown on the page are conservative because full attribution is not possible.

Portfolio totals as of today (seven live surfaces, iOS and Android combined): 363 downloads. Two weeks ago: about 200. Change: roughly +80 percent over two weeks. The largest single surface accounts for about half. Several others are still in the low twenties.

Last week's content output (one person, no shoot day)

YouTube: 23 videos, 3,800 views, 2.9 hours of watch time, 4 new subscribers.

TikTok: 6 videos, 621 views, 29 likes.

Instagram: 6 videos, 908 views, 3 new followers

Recording setup: real-device screen recordings are automated through an Android emulator on a Linux PC, with Claude driving the recording. Apps that use the camera through vision AI are the exception and still require manual capture.

What the pipeline does

Same as the earlier description. One selfie, two voice files from ordinary work calls, and one shipped APK produce a 40- to 60-second short. Script, avatar, narration, screen capture, assembly, and an upload kit (title, hashtags, description). Uploads to each platform are manual.

The pipeline is behind a private waitlist. Not for sale yet.

Caveats from the two weeks window

3.1 Volume alone does not compound. Twenty-three YouTube videos in a week moved four subscribers. Without a hook that lands in the first second, higher output does not change the ratio.

3.2 The 4 percent outbound click-through cited earlier needs a caveat. Each platform uses a different link tracker. Tying those clicks to specific downloads is logically weak. Treat the number as directional, not causal.

3.3 The pipeline is inexpensive, not free. Expensive services such as Hugging Face inference and ElevenLabs are avoided. Distribution APIs would be paid if this were offered as a service. For now the upload step is manual: a generated upload kit (title, hashtags, description) is posted to each platform by hand.

Questions

Four things I would like input on.

For non-English-country developers who shipped English apps: what actually moved your regional mix?

Has anyone seen a small landing page outperform social channels for downloads in the early weeks?

For solo makers running more than three products at once: how do you split time?

On the discovery-signal problem (about 560,000 new App Store apps in H1 2026, 2 percent download growth): does anyone see a community forming around this specific pain, or is it a personal reading?

Closing

The video pipeline has stabilized week over week, and output quality has improved with use. A separate post on how it is set up will follow if there is interest.

Team update: one developer who understood the project deeply joined. Two people now.

If you are stuck on marketing as a vibe coder, or have a working philosophy for what to do after an app ships, I would like to talk. Getting people to download the app is hard. If the project reaches some traction, I plan to share how the business side started openly. What I want to think about next is product distribution, with others.

If you're interested in my work, please check out the URL on my profile!

From Seoul.