"Is my host slow?" — answered with four graded probes. Run on demand. Keeps 12 runs of history so a host migration shows up as a chart trend, not a hunch.
Slow WordPress sites often aren't caused by a bad plugin or a theme — they're caused by a slow host. But "slow host" is hard to prove without numbers. The Hosting Benchmark gives you four timed measurements of the actual server hardware running your WordPress site: how fast the database responds, how much CPU the PHP runtime gets, how quickly files can be written and read, and how long a full HTTP round trip to your own homepage takes.
Results are graded A through D against published shared-hosting-vs-managed-hosting baselines — so you can see at a glance whether you're getting what you're paying for. Keeping 12 runs of history means a host migration or a gradual degradation in hosting quality shows up as a visible chart drop, not an anecdotal feeling that "the site feels slower lately."
The benchmark never runs automatically. It only executes when you click the run button — because the probes take 2–4 seconds and involve a file write in the uploads directory. Run it once per environment, after a host migration, or any time you suspect performance has changed.
Runs 150 simple indexed SELECT queries against your wp_options table and measures round trips per second. This measures the full database stack — connection overhead, query execution, and result transfer. Grade A is 900+ queries/sec (fast managed hosting, VPS), Grade D is under 250/sec (oversold shared hosting or an unindexed database). The most common bottleneck on slow WordPress sites is database latency, not PHP.
Runs 150,000 MD5 hashing operations and measures throughput in thousands of operations per second. This isolates pure PHP CPU performance — unaffected by database or network. Low scores here indicate either throttled CPU on a shared host, PHP process limits, or an unusually slow server — even when the database is fast. Grade A is 900k+ ops/sec; Grade D is under 250k ops/sec.
Writes a 256 KB temporary file to the uploads directory, reads it back, then deletes it — measuring the total round-trip latency in milliseconds. Filesystem speed matters for plugin file loading, image generation, cache file writing, and log writing. Grade A is under 15ms; Grade D is over 100ms. The file is cleaned up immediately after the test regardless of outcome.
Makes an HTTP GET request to your own homepage and measures Time to First Byte. This measures the full WordPress stack end-to-end: DNS, TCP, TLS, PHP execution, database queries, and response start. It's the closest approximation to what a real visitor experiences before any content loads. Grade A is under 400ms; Grade D is over 1800ms. Pairs with the What Changed? timeline to explain sudden TTFB increases.
Checks whether WordPress is using a persistent external object cache (Redis, Memcached) via wp_using_ext_object_cache(). Without a persistent object cache, WordPress hits the database for every cached value on every request — even if the data hasn't changed. Sites without object caching typically see 2–4× more database queries per request than sites with it.
Stores the last 12 benchmark runs and displays database speed as a bar chart with colour-coded grades. A sudden, sustained drop in the chart — especially after a specific date — is evidence of a genuine hosting degradation, a host migration, or a new plugin with a high database query count. Cross-reference the date with the What Changed? timeline to identify the trigger.
wp wphi benchmark --format=json) means you can include the benchmark in automated site audit pipelines without opening a browserIt adds a small, brief load — 150 extra database queries and one file write during the ~3 second probe window. On a low-traffic or staging site, this is imperceptible. On a very high-traffic production site, you may want to run it during a quiet period. It's roughly equivalent to one extra page load from a heavy plugin, and it never runs automatically.
Yes, especially on shared hosting — resource availability fluctuates as the server allocates to other customers. Run three to five benchmarks and look at the average rather than a single result. Sustained changes over multiple runs are meaningful; a single outlier is just variance. The history chart makes this easy to see at a glance.
Two common reasons: (1) your caching plugin serves cached pages to visitors faster than the loopback test sees, since the loopback sends a header that bypasses some caches; (2) your CDN serves cached assets to visitors from an edge location close to them, while the loopback measures your origin server directly. The loopback TTFB reflects your origin server performance — what happens on cache miss or for logged-in users.
Grade A (database 900+ q/s, PHP 900k+ kops/sec) is typical of managed WordPress hosts (Kinsta, WP Engine, Cloudways), dedicated servers, and well-configured VPS. Grade B–C is typical of mid-tier shared hosting with SSD storage. Grade D is typically oversold shared hosting with spinning-disk storage or heavily throttled CPU. The filesystem and loopback probes have similar tier distributions.
Yes — wp wphi benchmark runs all four probes and prints the grades to the terminal. Add --format=json for machine-readable output suitable for CI pipelines, automated site audits, or logging to an external monitoring system.
Four graded probes, history chart, object-cache detection, and WP-CLI support — free, forever, no account required.
Pro adds Root Cause Analysis, Request Profiler, Performance Baselines, and the broader Pro inspection suite — plus a weekly Health Digest.
Hosting Benchmark is fully free in Lite — alongside 12 other diagnostic modules, no account, no expiry.