We've tracked over 200 icon A/B tests run through Marteso, covering categories from productivity and games to health and finance. The data is surprisingly consistent — and often counterintuitive.

Why icons matter more than you think

In search results, your icon is often the only visual element a user sees before deciding whether to tap your listing. Store listing page views where the icon was the primary variable show an average 22% difference in click-through rate between the best and worst variants.

That 22% difference in CTR translates directly to installs — and indirectly to keyword ranking, since install velocity is an algorithm signal.

What consistently wins

  • High contrast backgrounds: Icons with strong contrast between foreground and background outperform low-contrast variants in 78% of tests
  • Faces and characters: Human faces and character-based icons show +15% average CTR lift in non-game categories
  • Single focal element: Icons with one clear focal element beat busy, multi-element designs in 71% of tests
  • Warm color palettes: Red, orange, and yellow backgrounds consistently outperform cool colors in Browse (not search) surfaces

What consistently loses

  • Text in icons: Any icon with text (including the app name) underperforms at small sizes
  • Gradients without contrast: Subtle gradient backgrounds get lost in the store grid
  • Category clichés: Icons that look identical to the top 3 apps in their category rarely beat those apps on CTR
  • White backgrounds: Performs poorly on iOS where white backgrounds blend into the store UI
The most impactful single change you can make: switch from a gradient to a solid, high-saturation background. This one change produces a measurable lift in 6 out of 10 tests.

How to run a valid test

Apple's Product Page Optimization tool requires a minimum of 90 days and enough traffic to achieve statistical significance. For most apps, this means at least 2,000 impressions per variant before the results are trustworthy.

Common mistakes that invalidate results: running a test during an unusual traffic period (holiday seasons, major update launch), testing too many variables at once, or stopping the test too early because one variant looks like it's winning.

Reading the data correctly

A lift in tap-through rate doesn't automatically mean the winning icon is better for your business. Look at downstream metrics: do users who installed via the winning icon have better Day-1 retention? Lower early churn? If the "better" icon attracts users who don't stick around, you may be optimizing for the wrong signal.