EWWW Image Optimizer keeps its own custom table at wp_ewwwio_images with one row per attachment file. Each row carries an id, the attachment_id, the original orig_size, the resulting image_size, a pending flag, and the level used for compression. The bulk run also uses a queue table at wp_ewwwio_queue for in-flight jobs. The default EWWW bulk optimizer screen renders progress as a counter and an in-page log, which works during the first run and becomes inscrutable once a site has tens of thousands of rows with a long tail of stuck or failed entries.

SleekView Kanban reads the same wp_ewwwio_images rows the bulk screen queries. Pick a derived ewww_state field that buckets rows by the pending flag, presence in the queue table, the level value, and any recorded error and every attachment becomes a card grouped under Queued, Running, Optimized, or Errored. Card fronts can show the file name, the original size, the resulting size, the savings ratio, and the level so a site manager can see exactly which files need attention without opening the bulk screen.

Dragging a card between columns updates the EWWW row through the plugin's own helper methods. A move from Errored back to Queued sets the pending flag to one and clears any recorded error. A move from Running back to Queued removes the row from the in-flight queue table cleanly. Any subscriber to the ewww_image_optimizer_post_optimization hook keeps firing on subsequent runs, so existing automations such as a CDN cache purge continue to work without changes.

EWWW Image Optimizer is one of the most flexible image optimization plugins for WordPress, which is also why it accumulates the longest tail of edge cases. The bulk screen is built for the happy path and the in-page log is unreadable once a site has thousands of rows with a mix of optimized, queued, and errored entries. Site managers either accept the long tail or block out an afternoon to chase down each error by hand.

The errors then quietly come back the following month, often from the same hosting configuration. A kanban view that reads and writes the same EWWW rows the bulk screen uses keeps the errored cohort visible, prioritized, and actionable. Queued images surface before publication.

Running images carry a live progress signal. Errored images cluster by message in a single column a developer can clear in an afternoon instead of letting them rot until next quarter's performance audit forces the issue.