Research · Auto repair
Three images, one second
For one in five homepages we checked, optimizing only three images could remove at least one second of image downloading on a 5 Mbps connection.
One in five could remove at least one second of image downloading at 5 Mbps.
The typical opportunity was small: a median of 0.14 MB, or about 0.2 seconds at the connection speed used here. The useful finding is the tail: 461 homepages had one to two seconds available to remove, and 535 had at least two seconds.
A focused fix, not a guaranteed loading-time reduction
Image transfer happens alongside other work in the browser. Caching, download order, scripts, and the image that becomes the main visible content all affect whether these savings shorten the final loading time.
This estimate still identifies a practical place to start. Three files are much easier to inspect and replace than an entire website.
How we got these numbers
We checked up to three images from accessible homepages associated with businesses categorized as auto repair. Each website was counted once using its latest homepage result.
Valid estimates were available for 4,985 homepages and covered 12,156 image records. We excluded 36 records where the estimated saving was negative or larger than the original image. Potential savings were added per homepage and converted at a fixed 5 Mbps using: megabytes multiplied by eight, divided by five.
An honest limitation: these are the businesses we chose to analyze, not a random sample of the industry. The result estimates image transfer time, not a guaranteed change to total page-load time or Largest Contentful Paint. Actual compression results also vary by image.
What to do about it
- Keep a backup. Save the original file before changing it.
- Resize the image. Match its dimensions to how large it appears on the page.
- Compress it. A tool such as TinyPNG can reduce common image formats.
- Consider a modern format. WebP can often reduce size while preserving quality.
- Replace and retest. Check the page visually and measure it again.