I'm a solo dev with a day job. My app data showed me where I was wrong.

I build apps alone, in Japan, after my full-time job. My main app is a kids drawing app for iPhone and iPad. It makes most of my indie revenue with zero ad spend.

For years I made decisions on gut feeling. This year I finally looked hard at the data, and almost every assumption I had was wrong. Here are the lessons, with real numbers.

1. One keyword update broke my conversion rate overnight

For 9 months my App Store conversion (page view → install) sat between 4.5% and 7.5%. Then on one day it dropped to 3.3–4.1% and stayed there.

I checked screenshots, the subtitle, and crashes. None of them explained it. The cause was my new keyword list. I had added broad, high-volume words like game, fun, cute, girls, boys, animals. They pulled my app into searches from people who were never going to install a kids drawing app.

I replaced them with narrow, intent-heavy words: tracing, toddler, preschool, crayon, sketchbook, kindergarten. Conversion went back up 38% within a week.

Lesson: every keyword should be something a person types when they're already looking for your app. Volume is a trap.

2. My paywall was everywhere, which was the problem

I assumed I needed more paywall exposure. The data said the opposite:

• The paywall showed on the home screen on almost every app open.
• That came to roughly 280,000 impressions a month for ~300 purchases, or 0.11% per impression.

I was asking for money at the front door, before the kid had drawn anything. Meanwhile, the screens where kids actually spent time (16+ minutes in the drawing canvas) had no ending screen at all. There was no "you finished!" moment, no share button, no well-timed upgrade prompt.

So I built a completion screen, a stroke-by-stroke replay of the drawing, and video export. The upgrade prompt now appears after the child has made something they're proud of.

Lesson: put the paywall where value is created, not where traffic is.

3. 58% of users never opened the main feature

A drawing app where more than half of the users never reached the drawing screen. The old menu was a grid of 8 options, and people spent 27 seconds on it before many of them left.

I redesigned home as a carousel and turned the old mini-games into level-based "drawing adventures." That release went live this month, so the results are still coming in.

Lesson: look at your funnel by screen, not just by install → purchase.

4. Paid ads didn't work at my scale, and that's fine

I tried Apple Search Ads with a tiny budget. At my maximum affordable bid, I got 3–5 impressions per day. The math never closed: the CPA I could afford was far below what the market charged.

Lesson: calculate your break-even CPA before spending. For many small apps, ASO and product fixes beat ads.

5. My biggest mistake: changing too many things at once

Here's the honest part. In 7 weeks I changed my price three times. At one point I changed the price and the keywords on the same day, so I couldn't tell which one moved the numbers. The data was saying "hold still." I didn't listen.

Lesson: one change, one measurement window. With noisy daily data, you need weeks, not days, to see a real effect.

What I'm doing now

Besides the app, I've started Doodle Cub: free browser games where the drawing is the game. You draw your car and race it, draw your snake's skin, or make one line work on both sides of the page. No download, no sign-up.

If you're a solo dev with a day job too: you don't need more features. You need to look at the numbers you already have.

What's the one metric that surprised you most in your own project? I'd love to hear it in the comments.