Reading Your Site Metrics: A Field Guide

Most hosting dashboards show a status badge, green, yellow, red, that tells you nothing useful. Real metrics tell a story. Four numbers, read together, will usually tell you what’s actually happening on your site: CPU, memory, latency, and request rate.

Request rate: the context

Start here. Request rate is how much traffic you’re getting. It’s the context for everything else, a slowdown during a traffic spike means something very different from a slowdown while traffic is flat. Always read the other three against this one.

Latency: the symptom

Latency is response time, how long your site takes to answer. It’s usually the symptom you notice first. Watch the percentiles, not just the average: a good average with a bad 95th percentile means some visitors are having a much worse time than the headline number suggests.

CPU and memory: the cause

When latency rises, CPU and memory tell you why. Latency up with CPU pegged points at heavy computation, an expensive query, an unoptimized plugin. Latency up with memory maxed points at a different problem, a leak, or simply not enough headroom. Latency up while both are calm points away from the server and toward something downstream.

Reading them together

The skill isn’t reading one number, it’s reading the shape of all four at once. “Response time rose at 2pm, request rate was flat, and CPU spiked with it” is a diagnosis. “The badge went yellow” is not.

The bottom line

If your host only shows you a traffic-light badge, you can’t actually diagnose anything. Real per-site metrics turn guesswork into a read. See what ours show at /help/understanding-dashboard.