100,000+ Images Repaired and Optimised for a Large WooCommerce Store

When we migrated the large WooCommerce site to new hosting, the media layer broke. Featured images failed to load, "can't find image" warnings appeared during regeneration, thumbnail sizes were missing. File counts and attachment records didn't match. Regeneration failed. Permission conflicts between server users caused unlink failures, lock file conflicts and logging errors.

In one affected directory alone, around 5,000 attachment records were out of sync with the files on disk; hundreds of large images exceeded 1–2 MB. This wasn't a simple thumbnail glitch. It was a database-to-filesystem mismatch at scale.

Site-wide, the image library needed attention: PNGs where JPG would do, oversized thumbnails, no consistent strategy. Fixing the broken layer was the first step; making the whole library smaller and faster was the second.

Work was done in two phases: repair, then optimisation.

Audit: Cross-checked attachment records against physical files in the affected directory. Identified mismatches between database and disk. Confirmed no missing source files.

Metadata repair: Repointed attachment records to valid -scaled files where originals were missing. Cleared wrong metadata. Validated image editor support (JPEG, PNG, WebP). That removed the regeneration warnings.

Batch regeneration: Built a custom batch script. Processed images in small groups. Logged progress, made execution resumable. Regenerated only the affected attachments. Normalised permissions so unlink and write operations succeeded.

Site-wide optimisation: Processed over 100,000 images across the installation; optimised more than half with major file size reductions. Regenerated thousands of thumbnails and resized them for faster mobile loading. Converted oversized PNGs to lighter JPGs where appropriate. In one key folder (~5,000 images), total size dropped from 4.56 GB to 970 MB; largest files from 5–8 MB to under 800 KB.

The WooCommerce media system was restored without re-importing products, re-uploading media or rebuilding the catalogue. The site serves correct thumbnail sizes, consistent product imagery, no regeneration errors. Database and filesystem are aligned. The media pipeline is stable and maintainable; the image library (2020–2024) is fully optimised.

What this means for the site: Faster page loading, better mobile performance, reduced bandwidth use, improved SEO from speed gains. Product pages stay fast while keeping good visual quality. Proper thumbnail sizes are served in grid and listing views instead of full-size images. Layout shift from missing thumbnails was removed; regeneration and permission errors were eliminated. The image library is in a good state for ongoing use.