A news aggregator moved SSR from a single Node region to edge functions for speed. TTFB dropped beautifully. Then -19% on breaking-news queries where minutes decide ranking.
The investigation traced a freshness gap. Their edge runtime read from a regional cache to stay fast — but that cache lagged the origin database by up to 90 seconds at the far edges. Googlebot, crawling from varied locations, sometimes hit an edge node serving a 90-second-old homepage missing the just-broken story. Fast, and behind.
Turns out edge SSR optimizes latency by moving compute away from your data — and freshness is the bill. For breaking news, 90 stale seconds is the entire window.
They kept edge rendering for evergreen pages and pinned breaking-news templates to a node region colocated with the database, accepting +40ms TTFB for live data.
Stale-homepage crawls: gone. Breaking-news queries recovered +14% in five weeks. Edge buys latency by spending freshness. Spend it on pages that don't change, not the ones that change by the minute.
Hydrate Diaries
@HydrateDiaries
The edge made it fast and the data made it stale
This post was published in the Telegram channel Hydrate Diaries. Subscribe here: @HydrateDiaries.