Optimizing Mobile App Icon Size to Reduce Traffic Consumption: Analysis and Results

1. New Challenge

The product line identified client‑side traffic consumption as a key performance issue and requested a detailed optimization report, which initially overwhelmed the QA team with unfamiliar terminology.

2. RD Analysis Report

The RD provided a breakdown of traffic sources, including APK downloads, app icons, web pages, splash screens, and CS communication, each contributing varying amounts of data usage.

3. Existing Problems with App Icons

Icons are frequently fetched due to high usage in common views, and current icon sizes (e.g., 200 px WebP) are larger than necessary, especially on low‑density devices.

4. Icon Optimization

The optimization replaces resolution‑based sizing with DPI‑based tiers. A linear fit between screen density and required icon size was derived, leading to revised size buckets (65 px, 90 px, 130 px, 200 px) mapped to DPI ranges.

After adjustment, low‑density phones see a reduction from 130 px to 90 px icons, saving traffic without noticeable UX loss.

5. Post‑Optimization Effects

Devices like the Asus Zenfone 6 (≈245 dpi) will fetch 90 px icons instead of 130 px, yielding measurable data savings over time.

6. Preliminary Analysis

The algorithm change is minor—only the image size calculation shifts from resolution to DPI—but raises concerns about potential UX impact on high‑DPI, low‑resolution devices.

7. Basic Concepts

Definitions of screen resolution, DPI/PPI, and DP are clarified, along with the Android formula for converting DP to actual pixels:

Actual display pixels = (DPI / 160) * dips

Icon DP values differ per density tier, as shown in the RD’s code snippet.

Android defines six generalized densities (ldpi, mdpi, hdpi, xhdpi, xxhdpi, xxxhdpi) ranging from ~120 dpi to ~640 dpi.

8. Data Collection

Screen resolution and DPI statistics were gathered from a major hardware review site, providing a basis for estimating the impact of the new algorithm.

9. Data Organization

Analysis of the collected data shows that certain screen sizes (e.g., 480×800 px, 540×960 px) would actually increase traffic under the new scheme, while low‑density devices benefit from reduced icon sizes.

10. Report Generation

The final report includes estimated traffic changes based on sampling 200 px, 130 px, 90 px, and 65 px WebP images.

11. Victory

After discussion with RD, the icon size algorithm was corrected, and the detailed report received positive feedback.

12. Revised Test Results

13. Summary

The two‑day effort transformed a vague problem into a data‑driven solution, teaching the team to confront unfamiliar challenges by researching, asking questions, and applying systematic analysis.