App Store Screenshots with GPT-Image-2 ($1.6k mrr in 36 days)

Your app store page is one of 4 conversion variables:

Marketing -> App Store Page -> Onboarding -> Paywall.

People focus on the onboarding and paywall. The product funnel starts at the app page! If your screenshots are bad, people won't download!

My app has a 65% page view to download.

This is a huge part of why I got to 1.6k mrr so fast.

I'll show you how to do this for my new app!

**The best part?

1. **With AI: This takes < 30 minutes!
2. Most app store pages are cooked. The effort to beat a competitor is tiny!This is all you need:
3. Pinterest account (FREE).
4. Figma account (FREE).
5. GPT-Image-2 (Free with a $20/m codex plan).

Step One: Research.

We'll be making a feminine manifestation app.

First: Look at other apps in the niche & spot patterns.

Credit: "Stella" & "Manifest Affirmations" Apps

What I notice:

• Pink & warm tones.
• Highly "aesthetic", images as a focus.
• Selling the outcome: not the product.

**Second: Browse Pinterest.

**This will give you broader context on your niche.

As well as asset ideas to use later.

Common Trends:

• Words: Magnetic, healthy, glowing, presence, love.
• Soft energy, powerful mantras.
• Save any images you really like:

The bottom left is a banger in itself.

Step Two: Figma MVP

The cheat code to getting cracked results from AI:

Is to make a mockup in figma first.

*If you prompt without any references you'll get bad results. Putting in a small amount of upfront effort has outsized ROI.*

1. **Templates:

**There's a lot of free templates on Figma. I use this one:

2. Fonts

There's two kinds of fonts everyone uses:

• The "premium" high end look: Instrument Serif.
• The modern, bold minimal look: SF Pro.

It's impossible to go wrong using either of these:

The other two tools are the gradient generator & liquid glass components.

I made these MVPs in just a few minutes.

Step Three: The AI Magic

Now, prompt GPT-Image-2 with our MVP references.

Make these screenshots more engaging and slightly more vibrant. Make UI elements subtly pop out, overlapping the phone border, giving a 3d effect. Output x separate images.

One recommendation here:

Keep the model effort on medium.

Play around and experiment of course, but for me, higher effort resulted in weird results: odd aspect ratios, funky text sizing etc.

These are my results:

1. Contrast. The UI elements are "3D", they pop out of the phone. This makes them engaging to the eye. If you look at popular apps, almost all do this.

1. Subtle vibrancy. Added colour aware glow and sparkle UI elements. These small upgrades have a subconscious effect on the user. The final professional finishing touches.

1. Image Filters. The model has boosted vibrancy and saturation. This can sometimes be overdone, so keep an eye on it. But here, it's clearly more engaging!NOTES:

Sometimes the AI will boost saturation a little too much, making images look plastic and AI-like. You can adjust the initial prompt, removing the "vibrancy". Or, ask to revise the outputted images!

I've kept it closer to the original as the MVPs turned out quite nice, but you can go crazy with this step, completely redesigning if you want to. The important bit is giving a "good enough" reference so the model knows what direction to go in.

Step Four: Final Touches

Apple requires very specific resolutions.

The image model won't output them in the right way.

We can either:

1. Manually edit in Figma (preferred). Takes 2 seconds.
2. Ask Codex or Claude to trim to the correct size.

TLDR:

1. Research App Store & Pinterest.

2. MVP in Figma.

3. AI Magic.

4. Make right size for Apple.