While reviewing the Stackra homepage, we noticed the logo image was an unoptimized PNG file. It weighed 350KB and measured 1024 pixels wide. On screen, it rendered inside a small navbar container at a fraction of that size. Visitors were downloading far more data than they needed to see the image.

Why image size matters for your business

Large images are one of the most common reasons websites load slowly. Google measures how quickly the main content of a page appears. This is called Largest Contentful Paint, or LCP. If your main visible element is a large image, it determines how fast your page feels. Google recommends LCP under 2.5 seconds. Pages that miss this threshold rank lower in search results. They also lose visitors. Research consistently shows that more than half of mobile users leave pages that take over 3 seconds to load.

What we changed

Two changes solved the problem. First, we converted the PNG to WebP format. WebP is a modern image format that all major browsers support. It delivers the same visual quality at a fraction of the file size. Second, we resized the image from 1024 pixels wide to 256 pixels wide. That is still sharp on high-density screens and appropriately sized for a navbar logo.

The result

The file dropped from 350KB to 2.4KB. That is a 99 percent reduction. The image looks identical to visitors. The difference is entirely in how fast the page delivers it. A logo is one of the first assets a browser requests. Cutting it from 350KB to 2.4KB removes a significant bottleneck on every page load.

What we found next: the favicon

After fixing the logo, we checked the other image assets on the page. The favicon was also a 1024x1024 PNG at 419KB. Browsers request the favicon on every page a visitor loads. It sits in the browser tab and is used for bookmarks and app installs on mobile. Despite rendering at 16 or 32 pixels, we were serving a file more than four times larger than the logo we had just fixed. The apple-touch-icon, which iPhones save when a user adds the site to their home screen, was the original 350KB logo file serving as a 180x180 image.

How we fixed the favicon and browser icons

We generated a proper set of correctly-sized icon files using ImageMagick, which resizes images without requiring extra software to be installed. The browser tab favicon became a 5.4KB ICO file containing 16x16 and 32x32 sizes, and a 1.8KB 32x32 PNG as a fallback. The apple-touch-icon was exported as a true 180x180 PNG at 8.6KB. The web app manifest, which controls what icon appears when a visitor installs the site as a mobile app, was updated with properly-sized 192x192 and 512x512 PNG files. Every declared size now matches the actual pixel dimensions of the file.

The combined result

The favicon went from 419KB to under 6KB. The apple-touch-icon went from 350KB to 8.6KB. The manifest icons are now correctly specified. These are not images that drive LCP, but they are requested on every page and on every return visit before cache warms. Fixing all of them removed unnecessary overhead that accumulates across many page views, especially for users on slower connections.

How to check your own images

Start with Google PageSpeed Insights. Paste your homepage URL and look at the Opportunities section. If images are slowing your page, they will appear here with the specific file and how much can be saved. The most common fix is converting PNG and JPEG files to WebP. For compression without any technical setup, TinyPNG handles it in your browser at no cost.

  • Google PageSpeed Insights ↗

    Check your page speed and see which images are flagged as oversized or unoptimized.

  • TinyPNG ↗

    Compress PNG and JPEG files to WebP format directly in your browser, no technical setup required.

What Stackra flags in your scan

When you run a Stackra scan, the system runs a full performance audit automatically. If a large image is slowing your LCP, the Action Plan will identify it. It will show the specific load time and explain how it affects your score. The finding connects directly to your Speed and Security pillar. You do not need to run a separate tool. The data is already in your report.