It's unclear if the tools used (Pingdom/Neustar) put cache-busting headers on their requests. It's likely that they do: such probes are more comparable/reproduc

It's unclear if the tools used (Pingdom/Neustar) put cache-busting headers on their requests. It's likely that they do: such probes are more comparable/reproduceable between runs, and more likely to give guidance to the app-designer/web-designer (as opposed to cache-runner). If so, of course CloudFlare would be slower in such tests: its largest benefit, caching unchanged resources for subsequent reloads, has been disabled. Also, it's hard to take performance hints from someone who splits such a short, simple blog post across 6 (!) pages. Six discretionary click-requiring page-loads is always worse than one, and is the easiest thing to fix if you're respecting my reading-time.