Purge by tag vs. by URL: a Madrid retailer kills a 9-minute stale window

Purge by tag vs. by URL: a Madrid retailer kills a 9-minute stale window
A fashion store with 80,000 SKUs ran URL purges. When a price changed, their script looped product, category, search, and homepage URLs — 14 calls per item, rate-limited to 200/sec.
A flash sale touched 6,000 products. The purge took 9 minutes; Madrid shoppers saw old prices the whole time.
They switched to cache-tag purging: every response carried a 'product-4471' tag, and one tagged purge invalidated every page referencing that SKU in a single API call.
The same 6,000-product update purged in 4 seconds.
URL purge is fine for a blog with flat paths. Tag purge wins the moment one entity appears on many URLs — products, authors, anything cross-referenced.
The number that mattered: 9 minutes down to 4 seconds.