Link prefetch vs. server-side cache warming: a Helsinki publisher beats cold cache

Link prefetch vs. server-side cache warming: a Helsinki publisher beats cold cache
A magazine published a big feature every Monday at 9am. Cold cache meant the first reader in each region triggered an origin miss; with traffic concentrated at launch, thousands hit a cold edge at once and origin latency spiked to 1,800ms.
They first tried browser prefetch hints — but those only help users already on the site, not the launch-minute flood arriving from social.
So they added server-side cache warming: a script fired the new URL through every major PoP two minutes before publish, pre-filling the edge.
Launch-minute origin requests dropped 96%. First-reader latency in every region held under 120ms.
Prefetch wins for predictable next-clicks within a session. Cache warming wins for a known asset facing a synchronized cold-start crowd.
The number that mattered: 96% fewer origin hits at launch.