Why you should test, not trust
Every host advertises speed and uptime. Advertising is not data. The honest way to evaluate a host is to measure it yourself — which is exactly the method behind every provider review and the benchmarks on this site. This guide teaches you that method in about 20 minutes.
The three things to measure:
- TTFB — how fast the server responds (the host's true signal)
- Load time / Core Web Vitals — what visitors experience
- Uptime — reliability over time
Test 1: TTFB from multiple locations
TTFB (time to first byte) is the single most host-specific metric — caching and front-end tricks barely touch it. A slow TTFB means a slow server, full stop.
Tools:
- WebPageTest — pick test locations (e.g. Mumbai, London, Virginia) and read the TTFB in the waterfall
- KeyCDN performance test — free, tests 14 locations at once
- curl (for techies):
curl -w "TTFB: %{time_starttransfer}s\n" -o /dev/null -s https://yourdomain.com
The numbers: under 200ms is excellent, under 400ms acceptable, above 600ms is a hosting problem. Our measured medians: Kinsta 98ms, Cloudways 124ms, Hostinger 142ms, SiteGround 185ms — see the benchmarks.
Test 2: Load time and Core Web Vitals
TTFB tells you about the server; load time and Core Web Vitals tell you what visitors feel:
- PageSpeed Insights (pagespeed.web.dev) — LCP, INP, CLS scores with lab data
- WebPageTest — full waterfall: where every millisecond goes
Run 3 tests and take the median (spikes skew averages). Watch:
- LCP under 2.5s — the hero/content load
- INP under 200ms — interactivity
- CLS under 0.1 — layout stability
If TTFB is fast but LCP is slow, the bottleneck is front-end (images, CSS, scripts) — fix that with our speed guide. If TTFB is slow, no front-end work saves you.
Test 3: Uptime over time
One uptime check is a snapshot. Real reliability needs time:
- UptimeRobot (free) — checks every minute, 2+ locations
- Better Stack / Pingdom — history and SLA-style reports
- Run for 30 days minimum before trusting a host
Track both hard downtime and slow responses — a site that responds in 30 seconds is down in practice. Full guide: WordPress uptime.
The complete method (what we do)
- Create a fresh WordPress site on the host (no optimized theme)
- Run TTFB tests from 12 global locations, 3 runs each, median
- Run PageSpeed + WebPageTest for Core Web Vitals
- Monitor uptime continuously for 30+ days
- Repeat after a traffic spike to catch throttling
This is the exact methodology published on our methodology page — you can reproduce it on any host you're considering.
Reading the results
- TTFB sub-200ms + uptime 99.9%+ → the host is genuinely good. Compare against the best WordPress hosting list.
- TTFB 400ms+ → look elsewhere or fix your own setup first (region, PHP version — see PHP versions).
- Slow only from one region → consider a CDN or a nearer data center rather than switching hosts.
Frequently asked questions
What is a good TTFB? Under 200ms is excellent (Kinsta, Cloudways, Hostinger territory); under 400ms is acceptable; above 600ms indicates a slow host or misconfiguration.
Is TTFB the most important hosting metric? Yes — it's the most host-specific. A slow TTFB cannot be fixed by caching plugins; only the host (or your server config) can.
How long should I test hosting before trusting it? 30 days minimum for uptime. Speed can be measured in a day, but reliability needs a month of continuous monitoring.
Do I need to pay for speed testing tools? No — WebPageTest, PageSpeed Insights, UptimeRobot, and KeyCDN's free test cover the whole method.
How do I test my WordPress site's TTFB?
Use WebPageTest (choose locations and read TTFB in the waterfall), KeyCDN's free multi-location test, or curl -w from your terminal.