Web hosting for furniture stores — Furniture retail: big images, big catalogues, uncached calculators

Furniture is photographed in room sets, which means wide images with a lot of detail, several per product, and a customer who wants to look closely before spending four figures. That is a page-weight problem before it is anything else.

Then there is the calculator. Delivery cost and lead time depend on a postcode and a product, so the answer is computed per visitor and cannot be served from cache. It sits on the page that has to convert, and it is the part most stores never measure.

Separating those two problems is most of the work. Here is how to take each reading, what to change, and how to tell that the change helped.

Furniture catalogues carry a lot of attributes: dimensions, materials, finishes, fabric options. Faceted filtering across those attributes generates queries that get expensive quickly, and a filtered category page is usually uncacheable because the combination is unique to that visitor.

Redis object caching is the lever here — it keeps repeated term and meta lookups in memory rather than returning to the database for each one. NVMe storage does the rest by making the round-trips that remain genuinely quick. Measure by filtering down to a narrow combination and timing the response, before and after.

Clone to staging before you touch the image pipeline or swap a filtering plugin. Measure on the clone, push when the number moves, and keep the before-and-after somewhere you can find it in six months when someone asks why the site is configured this way.

Retest one change at a time and take the median of three lab runs — a single run tells you about the test server's afternoon, not about your site. Then let the field data confirm it over the following weeks, because that is the version search engines read.