Most WordPress hosting comparisons show you synthetic benchmarks and uptime percentages that tell you nothing about how your site will actually perform. I've watched hundreds of migrations where a host looked great on paper but fell apart under plugin load or produced wildly inconsistent response times.
The tests that matter aren't the ones vendors advertise. They're the ones that expose caching gaps, reveal database bottlenecks, and show whether a host can handle WooCommerce with thirty plugins without grinding to a halt. Here's how to run six practical speed tests that predict real-world WordPress performance.
Test 1: Cold-start TTFB with cache disabled
Time to First Byte tells you how fast the server processes a PHP request before any caching kicks in. This is your baseline.
Disable all caching layers — page cache, object cache, CDN — and measure TTFB for the homepage and a typical post. Use curl with timing:
curl -o /dev/null -s -w "TTFB: %{time_starttransfer}s\nTotal: %{time_total}s\n" \
-H "Cache-Control: no-cache" \
https://yoursite.com/
Run it ten times and average the results. A good shared host delivers 400-800ms. Managed WordPress platforms should hit 200-500ms. VPS setups with proper tuning often land under 300ms.
What ruins TTFB? Overloaded MySQL instances, ancient PHP versions, too many simultaneous sites on the same Apache worker pool. If cold TTFB exceeds one second, the host's infrastructure can't keep up.
Test 2: Cache hit ratio under repeated load
Page caching should serve most visits from memory or disk without touching PHP. The gap between cached and uncached response times reveals how well the host implements caching.
Measure the same URL three ways: cold (cache miss), warm (second request), and after fifty rapid requests. Use Apache Bench:
# Clear cache first via plugin or server control
ab -n 50 -c 5 https://yoursite.com/sample-post/
Check the "Time per request" line. Cached responses should be 10-30ms. If warm requests still take 200ms+, the host's page cache isn't working or expires too aggressively.
I've seen hosts advertise "advanced caching" but configure it to expire every five minutes or exclude logged-in users so poorly that cart pages never cache. The hit ratio test exposes that immediately.
Test 3: Plugin-heavy page with WooCommerce
WordPress sites in production run fifteen to forty plugins. A host that flies on a bare install but chokes under plugin load is useless.
Install a realistic plugin stack: WooCommerce, a page builder (Elementor or Divi), an SEO plugin, a form plugin, and a couple of typical utilities. Create a product page with images and variations.
Measure TTFB and full page load for that product page with cache disabled. Compare it to the baseline homepage test. If the plugin-heavy page is more than 2× slower, the host's PHP execution or database performance can't handle realistic workloads.
Shared hosts often throttle PHP workers or database queries when usage climbs. That's fine for brochure sites but fatal for shops or membership sites.
Test 4: Database query performance under concurrency
WordPress hits the database constantly — posts, options, user data, transients. Slow queries stack up fast when you have concurrent visitors.
Enable the Query Monitor plugin and load a typical page. Check the query count and total query time. A well-optimized WordPress install runs 20-40 queries per page in under 50ms total. If you see 100+ queries or database time over 200ms, something's wrong — either the host's MySQL is overloaded or your plugins are poorly coded.
Now simulate five concurrent visitors. Use siege:
siege -c 5 -t 30S https://yoursite.com/sample-post/
Watch response times. If they double or triple under light concurrency, the database pool is undersized or shared with too many other accounts.
Managed WordPress hosts typically isolate database resources better than budget shared plans. VPS setups give you full control but require tuning MySQL settings yourself.
Test 5: Object cache impact on admin dashboard
The WordPress admin is often slow because it can't rely on page caching. Object caching (Redis or Memcached) speeds up repeated database queries for options, post meta, and transients.
Without object cache, load wp-admin/edit.php and note the load time. Install Redis Object Cache or a similar plugin, verify it's connected, then reload. Admin pages should feel noticeably snappier — often 30-50% faster.
If enabling object cache makes no difference, the host either doesn't provide a persistent cache daemon or the network latency to reach it is too high. Some hosts advertise Redis but run it on a separate box with poor connectivity.
Object cache matters most for sites with lots of taxonomies, custom fields, or user roles — basically any site beyond a simple blog.
Test 6: Geographic latency and CDN integration
Even a fast host looks slow if most visitors are on another continent. Test from multiple locations using a tool like WebPageTest or a set of VPS instances in different regions.
Measure TTFB and full load time from at least three locations: US East, Europe, and Asia-Pacific. Without a CDN, expect TTFB to increase by 100-300ms per region due to network distance.
Now enable the host's included CDN (if any) or connect Cloudflare. Re-test. CDN should reduce TTFB variance across regions to under 100ms because static assets and cached pages serve from edge locations.
Hosts that bundle a CDN sometimes configure it poorly — cache TTLs too short, exclusion rules too broad, or only caching images while skipping HTML. Your tests will show whether the CDN is actually doing work.
What the tests reveal about hosting tiers
Shared hosting typically scores okay on cached requests but falls apart on cold TTFB and concurrency. That's expected — you're on a box with dozens of other sites. Fine for low-traffic blogs, not for anything business-critical.
Managed WordPress platforms optimize the whole stack for WordPress specifically. They usually deliver strong TTFB, good caching, and handle plugin load reasonably well. The tradeoff is less control and higher cost.
VPS or cloud hosting gives you the best raw performance if you configure it properly. But you're responsible for PHP-FPM tuning, MySQL optimization, and setting up Redis yourself. If you don't know how to do that, managed is safer.
What should you avoid? Hosts where cold TTFB exceeds one second, where enabling cache makes no measurable difference, or where light concurrency causes response times to spike. Those are signs of oversold infrastructure.
How to interpret your results
You're not looking for perfection. You're looking for consistency and reasonable performance under realistic load.
Good indicators: - Cold TTFB under 600ms on shared, under 400ms on managed or VPS - Cache hit serving pages in under 50ms - Plugin-heavy pages less than 2× slower than the baseline - Concurrent load adding less than 50% to response time - Object cache cutting admin load time by at least 20% - Geographic TTFB variance under 100ms with CDN enabled
Red flags: - Any metric more than 3× worse than the good threshold - Response times that spike unpredictably (suggests resource contention) - Cache that barely helps or expires too fast - Database queries piling up under light concurrency
If a host fails more than two of these tests, it's not going to support a production WordPress site well.
Realistic test scenarios matter more than marketing claims
I've migrated sites off hosts that advertised "blazing speed" and "optimized for WordPress" but couldn't handle a WooCommerce store with twenty products. The marketing looked great. The actual performance didn't.
Run these six tests on your current host and any host you're evaluating. Skip the vendor's demo site and test your actual site or a realistic clone. The results will tell you more than any comparison chart.
Speed matters for conversions, SEO, and user experience. But only if it's real speed under real load, not synthetic benchmarks on an empty install.
FAQ
Q: Can I run these tests on a host before buying?
Most hosts offer 30-day money-back guarantees. Spin up a trial site, install your typical plugin stack, and run all six tests within the refund window.
Q: What if my site is slow but the host's test results look fine?
Often the problem is plugin bloat, unoptimized images, or missing object cache. Profile your site with Query Monitor first before blaming the host.
Q: Are managed WordPress hosts worth the extra cost?
If you don't want to manage caching, PHP versions, and server config yourself, yes. If you know your way around a VPS, you can match or beat managed performance for less money.
Q: How often should I re-test my host?
When traffic grows significantly, after a migration, or if you notice performance degrading. Infrastructure can degrade as hosts oversell capacity.
Start with cold TTFB and caching behavior
If you only run two tests, make them cold TTFB and cache hit ratio. Those two metrics predict 80% of real-world WordPress performance. Everything else is refinement.
WordPress hosting providers differ more than their marketing suggests. The hosts that perform well under realistic testing aren't always the ones with the loudest claims or the biggest ad budgets. Test what matters, not what they want you to measure.