How We Test Hosting
This page describes how HostLab Reviews produces scores. We designed the protocol so another operator could reproduce the structure of the test — identical controls, transparent weights, and clear limitations.
1. Comparison set
For 2026 we evaluated five providers that map to real reader decisions: transparent shared hosting (SupportHost), managed WordPress (Kinsta), privacy-oriented VPS/RDP (RDP Monster), mainstream shared/cloud (SiteGround), and trading-oriented low-latency VPS (QuantVPS). This is a focused lab set, not a claim that no other host can be excellent.
2. Environment control
- Identical WordPress 6.4 installs on each provider’s recommended plan for its category
- Same theme, plugin set, and 8.2 MB synthetic content page for load tests
- Vendor-default caching for “out of the box” reality; optimized passes labeled separately when used
- HTTPS everywhere; free SSL where the host provides it
3. Uptime
UptimeRobot HTTPS checks every 60 seconds for 30 consecutive days. We report percentage uptime and note multi-minute incidents. Planned maintenance still affects measured availability unless we could classify a distinct maintenance response.
4. Speed
TTFB and full-page load sampled every 6 hours from US-East, EU-West, and APAC (GTmetrix). Average page load is the primary consumer number. Specialty metrics (for example exchange latency on trading VPS) are reported in addition when the product is not primarily a public CMS host.
5. Support
Three blind tickets per host: billing clarification, a technical misconfiguration we introduced, and a migration feasibility question. We score time-to-first-response and whether the reply solved the issue. We never identify as reviewers in tickets.
6. Scoring weights
| Factor | Weight | What good looks like |
|---|---|---|
| Speed | 30% | Low, stable multi-region load times |
| Uptime | 25% | ≥99.9% with short incidents |
| Support | 20% | Fast, competent, complete answers |
| Value | 15% | Performance per dollar + renewal honesty |
| Features | 10% | Backups, staging, CDN, category fit |
7. What we refuse
- Free extended trials ordinary customers cannot get
- Vendor previews of scores before publication
- Selling #1 placement
- Hiding material renewal price increases in value scoring
8. Limitations
Thirty days catches instability patterns; it does not prove multi-year reliability. Shared hosts can be noisy-neighbor sensitive. VPS results depend on plan size. Trading VPS products optimize for market connectivity, not general web CMS speed — we judge them against their stated purpose. Prices change; we timestamp entries and correct material errors when verified.
Protocol questions: hello@hostlabreviews.com.