How do you improve App Store screenshots with AI? The exact prompt.

When working on screenshot strips, I want competitive insights instead of vague “design feedback.”

My exact process has three steps:

1. Upload my screenshots as a strip (one image with all screenshots in order)
2. Add 10 reference strips from top apps in the category
3. Use this prompt:

Task:
I'll upload my new screenshot strip plus a folder of 10 reference
screenshot strips from top competitors.

Output format:
Give me a Markdown table with three columns:
- Change to make (one short sentence)
- How many leaders already do it (X / Y form)
- Why it matters (one short sentence)

Rules:
- Use exact app titles
- Keep language simple
- No row should run longer than one sentence
- Flag anything done by fewer than 5 / Y apps as experimental
  in the "Why" column
- End with a one-line summary
- If you have extra points to tell, be proactive, don't make me
  miss important things

What this gives me:

• Feedback backed by data instead of design opinions
• Priority ranking based on what successful apps actually do
• Clear next steps

Part of an example output: “Add download count” (8/10 do it), because social proof drives downloads.

How I act on each row

1. I check the count. I look at the reference strips and confirm the count myself before I act on a row.
2. I start with the highest counts. The changes more leaders already make come first, and anything done by fewer than 5 of the 10 carries the experimental flag in the “Why” column.
3. I check the claim. If the change puts a claim on the screenshot, like a download count, I only show a number the app can back up.
4. I test it. Each row is a test, not a verdict, and I judge it on the app’s own store page conversion.

I work this way because screenshot work should start from what category leaders do and end with what my own test shows, not with what looks pretty.