How it works
How a server-side web optimizer works
The docs tell you which directives to set; the blog covers what to fix on a given stack. These are the explainers in between — how the optimization actually works under the hood. The techniques trace back to the open-source mod_pagespeed and ngx_pagespeed projects, which We-Amp helped build and maintain; ModPageSpeed 2.0 and mod_pagespeed 1.15 carry them forward.
How a Server Optimizer Avoids Adding Latency
How ModPageSpeed optimizes images, CSS, and JS without slowing the request: the second-request problem, and how 2.0 moves encoding off the request path.
Read itHow the Metadata Cache Avoids Re-Optimizing
How ModPageSpeed avoids re-optimizing on every request: a metadata cache keyed by the output URL maps to the answer and skips rewrites that don't shrink.
Read itHow CSS Minification and URL Rewriting Work
How ModPageSpeed parses CSS into a syntax tree to minify, rewrite url(), and flatten @import safely — and why regex rewriters corrupt stylesheets.
Read itFor the full list of what gets rewritten, see the feature list; for fixes by platform, see the blog.
Run it: ModPageSpeed 2.0 docs for the out-of-process worker, or mod_pagespeed 1.15 docs for the in-process module on Apache, nginx, and IIS.
Measuring your own pages? Test a page and map every failing audit to its fix, or read what LCP, CLS, and INP mean.