A directory is a search engine wearing a theme.

A directory looks like a content site and behaves like a database application. Every visitor arrives and immediately starts combining filters — category, location, price band, date posted — and each combination is a query that has never been run in that exact shape before.

That is why page caching alone disappoints here. The homepage gets fast, the about page gets fast, and the one screen everybody actually uses stays exactly as slow as it was, because it cannot be pre-built.

What moves the needle is the layer underneath: an object cache holding the repeated taxonomy and meta lookups, short-lived caching on the common searches, and a database that gets swept so the query planner is not fighting years of accumulated cruft. The filters start feeling instant, which is the only feature this kind of site really has.

Using xSpeed, my WordPress directory site is slow on filtered browsing. Benchmark a filtered search first, then connect Redis object caching, enable search results caching with a short lifetime, cache static pages aggressively, and schedule a weekly database cleanup with table optimisation. Re-run the same filtered search and show me the before/after.

What happens: Your AI reports the query time on a real filtered search before and after — the number that decides whether people keep filtering.