Disclosure: this is a practical how-to, not a comparison of specific hosts. If you want the actual scripts and configs we run, see our published test kit — this article is about what to do with them (or your own equivalent setup) on a host you're personally evaluating.
You don't need our exact repository to sanity-check a hosting review — including ours. A handful of free tools, run consistently, will tell you more about a candidate host than any published benchmark table, because you're testing the specific stack and content you actually plan to run.
What You Need Before You Start
- A cold-cache and cached-cache testing tool. WebPageTest lets you pick a test location, connection profile, and run count — use at least 3 runs and take the median, not the first result.
- A command-line TTFB check.
curl -o /dev/null -s -w "%{time_starttransfer}\n"against the same URL, repeated 5-10 times at different times of day, gives you a real distribution instead of a single snapshot. - A load-testing tool for concurrent-session claims. k6 (free, open source) or Apache Bench for a quick-and-dirty version — this is the step most readers skip, and it's the one that actually separates hosts under real traffic.
- Query Monitor installed on the test site itself, to see database query count and load time server-side rather than inferring it from the outside.
The Protocol, Condensed
- Use the same test content on every host. Comparing a bare WordPress install on Host A against a plugin-loaded site on Host B tells you nothing. Export/import the same demo site (or use a realistic stand-in with a similar plugin count to what you'll actually run) before testing.
- Test from a location close to your actual audience, not the host's data center. A host that looks fast from a test node next to its own server can still be slow for your real visitors if there's no CDN in front of it.
- Run cold and warm passes separately. First-visit (cache-empty) and repeat-visit (cache-warm) numbers can differ by 2-3x on the same host; report both, not whichever looks better.
- Test at more than one time of day. Shared and budget VPS hosting can show meaningfully different TTFB during that host's peak load hours versus off-peak — a single morning test can miss this entirely.
- Load-test before you commit, not after. A host that's fast for one visitor and falls over at 50 concurrent sessions is a different product than the marketing page suggests.
Where DIY Tests and Our Numbers Will Legitimately Differ
Even a careful reproduction won't match our published numbers exactly, and that's expected, not a red flag. Hosting infrastructure changes over time (a provider can migrate data centers or change default caching between when we tested and when you do), your test content and plugin list won't be byte-identical to ours, and network path variance between your test location and ours is real. Use our numbers as a starting point and directional comparison, not a guarantee of what you'll personally see.
DIY Tool Comparison
| Tool | Measures | Cost | Best for |
|---|---|---|---|
| WebPageTest | Full waterfall, Core Web Vitals, multi-location | Free | Realistic cold/warm-cache comparison |
| curl (CLI) | Raw TTFB, repeatable | Free | Time-of-day variance testing |
| k6 | Concurrent-session load behavior | Free (open source) | Testing under-load claims |
| Query Monitor | Server-side query count, load time, hooks | Free (plugin) | Diagnosing why a host is slow, not just how slow |
FAQ
Do I need our exact test site to get useful results?
No — using your own realistic content is arguably more useful, since it tells you about the host's fit for what you're actually running, not our reference case.
How many runs is “enough”?
At minimum 3 for a quick check; 5-10 spread across different times of day if you're making a real purchasing decision, since single-run numbers are the most common source of misleading benchmark claims (ours included, if we ever fell into that trap).
Is a load test really necessary for a small site?
If you never expect a traffic spike, it's lower priority. If you run anything affiliate, e-commerce, or content that could go viral or hit a seasonal spike, it's the single most informative test you can run and the easiest one to skip.
Verdict
Reproducing a hosting benchmark yourself takes an afternoon and entirely free tools — and it's worth doing before trusting any review, including this site's, at face value. The goal isn't to catch someone lying; it's that your specific plugin stack, traffic pattern, and audience location can shift the real-world result away from any single published number.
