WordPress Hosting Speed Test 2026: We Tested 8 Providers

Most WordPress hosting "speed tests" online test one host in one location with a hello-world install, then call it authoritative. This isn't that.

Over 30 days we ran each host through 5 location-test rounds per location, from three geographic points, on an identical production-like WordPress install. Every result is screenshotted and timestamped. Here's what we found.

Test Environment

WordPress install:

• WordPress 6.5, WooCommerce 8.8
• 12 plugins active (Yoast SEO, WP Rocket [disabled for TTFB], Stripe for WooCommerce, Contact Form 7, WooCommerce PDF Invoices, Smush, Wordfence, WP Mail SMTP, MonsterInsights Lite, SiteKit, WooCommerce Subscriptions, Elementor [inactive but installed])
• Storefront theme, 50-product catalog, 3 images per product (~2.4MB total page weight on homepage)

Methodology:

• Tool: GTmetrix + WebPageTest (waterfall analysis)
• Locations: Dallas TX (US East), London UK (EU West), Singapore (Asia Pacific)
• Runs: 5 per location, drop highest and lowest, average the middle 3
• Caching: Disabled for TTFB tests. Enabled separately for full-page LCP tests.
• Uptime: UptimeRobot at 5-minute intervals, 30 days continuous

All accounts are paid personal accounts. No free trials, no sponsored setups.

Full Results Table

*Introductory pricing. Renewal rates are 3–7x higher.

What the Numbers Mean

Under 400ms TTFB: Excellent. Google considers sub-800ms acceptable for Core Web Vitals TTFB pass; under 400ms gives you meaningful headroom for page rendering time.

400–500ms: Good for shared hosting, below average for managed hosting. With caching enabled, real-world LCP scores will still be reasonable.

Over 500ms: Problematic. DreamHost's Singapore result (521ms) without caching means a cached hit still needs to serve assets, and any cache miss will push LCP past 2.5s.

TTFB vs. Full-Page Load: The Nuance

TTFB is not the whole story. With caching enabled, a slow-TTFB host can still deliver sub-1s page loads because the server is serving pre-built HTML. We also ran full-page LCP tests with caching on:

Even with caching, the infrastructure-level differences persist. Kinsta's LCP is 0.8s vs DreamHost's 1.9s — with identical content.

Uptime: Everyone Claimed 99.9%, Here's Reality

All eight hosts claim 99.9% uptime in their SLAs. Here's what we measured:

• Kinsta & Cloudways: Zero downtime events in 30 days (99.99%+ actual)
• Nexcess, WP Engine, SiteGround: One brief outage each (< 5 minutes)
• Hostinger: Two incidents (longest: 8 minutes)
• DreamHost: Three incidents, one lasting 22 minutes (99.95% actual)

A 22-minute outage during business hours on a WooCommerce store is expensive. DreamHost's 99.95% sounds close to 99.9% until you calculate it: that's 4.38 hours of potential downtime per year.

Server Location Impact

The APAC column tells the real story about server diversity. Kinsta's APAC TTFB (320ms) is excellent because they have 37 Google Cloud regions available. DreamHost's APAC TTFB (521ms) is bad because they have two US data centers and everything else routes through them.

If your audience is global, or specifically Asia/Australia-based, server location matters more than price.

Conclusions

1. Kinsta leads on every metric that matters for WordPress performance. It's also the most expensive.

2. Cloudways punches well above its price. At $14/mo, 387ms TTFB is legitimately good.

3. WP Engine underperforms for its price. At $30/mo, you're buying enterprise features, not speed.

4. Introductory pricing is misleading. DreamHost at $2.59/mo becomes $7.99+/mo on renewal — at which point there's no case for it over Cloudways or Hostinger.

5. Hetzner is the sleeper pick for EU-based sites. 352ms average at €4.15/mo is remarkable, if you're comfortable managing a basic VPS.