An optimization plugin cannot fix an unknown delivery model. WordPress core, the active theme, and plugins may all register image sizes. Before adding another layer, identify which sizes exist, which templates request them, and whether the browser selects an appropriate candidate.

The decision this guide supports

Remove waste and close genuine candidate gaps without generating an uncontrolled library of derivatives.

Repeatable method

1. Inventory registered dimensions

Record core Media settings and image sizes registered by the theme or plugins. Distinguish cropped from proportional sizes.

2. Sample uploaded families

Choose recent large uploads and list every generated derivative with dimensions and bytes. Note missing files and legacy sizes from inactive themes.

3. Map sizes to CSS slots

For home, archive, article, and tool templates, record rendered slot widths and the currentSrc at representative viewports.

4. Change one layer at a time

If a size is never used, investigate before removal. If a gap causes oversized delivery, add one justified candidate and regenerate only with a backup and rollback plan.

Inventory one template before changing sizes

Pick a single template, such as a post card or article header, and audit it with one known media item. This fixture helps separate registered WordPress sizes from the file the browser actually selects. It does not establish an ideal size until you record real layout and delivery conditions.

Inventory field Illustrative record
Template / slot article-card / featured image
CSS slot width 360 CSS px at the tested viewport
Viewport / DPR 1280 px / 2×; replace with actual
Registered candidates Names, dimensions, crop mode, and generated files
Markup Observed src, srcset, and sizes
Delivered file Observed currentSrc and transferred bytes

Repeat for materially different breakpoints or templates. Do not delete old sizes until a backup and rollback path exist, and do not infer selection solely from the Media Library thumbnail.

What to record

Record WordPress version, active theme and plugins, registered size names, crop behavior, sampled attachments, derivative counts and bytes, template slot widths, selected candidates, and backup location.

Do not turn a lab result into a universal rule

A result belongs to the source file, encoder, settings, browser, viewport, cache state, and date that produced it. Repeat the test on representative files before changing a production workflow.

Common mistakes

  • Installing several optimizers before measuring
  • Deleting old derivatives without confirming references
  • Regenerating a large library without a backup
  • Treating every generated file as wasted disk space

Verification checklist

Recheck representative pages, broken-image logs, media editing, responsive candidates, disk usage, and backup restore instructions. Confirm no plugin rewrites markup twice or converts files that another layer already handles.

Primary references

Related lab notes

About this guide

We distinguish documented platform behavior from our workflow recommendations. See the testing methodology, editorial policy, and corrections log.