Thirty photos per car, and the stock changed this morning.

Sofia’s dealership lists around ninety used vehicles at a time. Each listing carries thirty photographs, a spec table, and a finance calculator — and buyers open four dealers’ listings for the same model side by side before deciding whose forecourt to visit on Saturday.

That made the site both heavy and volatile: enormous galleries that had to load fast, over inventory that changed several times a week as cars sold and arrived. Caching helped the speed and, configured naively, would have made the second problem worse.

The answer was a short cache lifetime on listings with auto-purge on edit, so a sold car disappears the moment it is marked sold, while the galleries themselves are converted, lazy-loaded, and served from static HTML. Fast enough to win the comparison, accurate enough to be worth the drive.

Using xSpeed, optimise my dealership site: convert and lazy-load the vehicle galleries, enable page caching with auto-purge so a sold car disappears from the listings immediately, set a short cache lifetime on vehicle listings and a long one on static pages, and connect the object cache for filtered browsing. Benchmark a vehicle listing page before and after.

What happens: Your AI reports the new listing-page weight and load time, with sold vehicles purging from cache the moment their status changes.