Alerting on time-to-first-byte vs. full page load time — which is the better health signal?

Alerting on time-to-first-byte vs. full page load time — which is the better health signal?

Q: For performance alerting, should I watch TTFB or total load time?

A: Use TTFB for server-health alerting and full load time for user-experience tracking — they're different signals. Time-to-first-byte measures how fast your server starts responding; a rising TTFB almost always means a backend problem (slow database, overloaded app server) and makes a clean, low-noise alert.

Full page load time includes images, fonts, and third-party scripts. It's what users feel, but it's noisy — a slow ad network or analytics tag can spike it without anything being wrong on your side, so it makes a bad alert trigger.

So: alert on TTFB (tight threshold, reflects your infrastructure), and chart full load time as a trend you review rather than get paged on. The follow-up: set the TTFB threshold from your own baseline plus a margin, not a generic round number — "normal" varies wildly by stack.

Got a question? Drop it in the comments.