The hidden cost of shipping unoptimized images
Putting images on the web is easy. Putting them on the web well is another story. Format choice, compression level, DPI, and dozens of other settings are easy to skip when you're handling everything by hand. The upside is that there's no shortage of tools and services that will handle these decisions for you. The downside is that the sheer number of options makes it hard to know which one fits your workflow.
That's the gap images.tooling.report is meant to fill. The site evaluates what the authors consider the most important features for shipping images in production, and compares popular software and services across paid and free, hosted and self-hosted options.
The motivation is straightforward: images account for a large share of bytes transferred on the web, and the data shows the impact of ignoring that fact. Unoptimized images make up nearly 75% of a page's total size. In a study of millions of real user sessions, pages that converted had 38% fewer images than pages that didn't. That doesn't mean ripping out all your <img> tags—but it does mean making sure what you send is optimized for every device that might request it.
What gets tested
Optimizing images is more than clicking "Save for Web," and the testing on images.tooling.report reflects that. The basics are covered—like compression levels and long-lived cache headers—but the checks go further into advanced territory.
For instance, the site looks at support for Save-Data, ECT, and other client hints. It also checks for automated psychovisual analysis: tools that use data models to repeatedly compress an image until it's as small as possible without perceptible quality loss. If a tool can squeeze out every non-visual bit before sending the file down the wire, it scores well.
No single tool or service scored perfectly across every test, so the results are broken into categories to make comparisons more relevant:
- CMS plugins cover packages for platforms like WordPress that are easy to integrate into an existing content workflow.
- Site builders compare services you can use to construct a full website.
- Self-hosted options target developers comfortable cloning a git repo or running their own Docker image in production.
- CDNs include both classic content delivery networks and services that proxy or host images at the edge.
The definitions are intentionally loose—they exist to make apples-to-apples comparisons easier, not to be strict taxonomies.
A living benchmark
If your image optimizer isn't listed, or if you think a test is missing, the team wants to hear about it. Both the list of tools and the tests themselves are meant to evolve. As the site states, it's a quick reference for the current state of image tooling—useful whether you're already optimizing or just starting to look at what the options can do for the people visiting your site.
