How we test

Hosting is the one corner of the WordPress economy where marketing claims are checkable in milliseconds. That is the whole premise of this site: instead of rewriting vendor feature pages, we put identical WordPress sites on the hosts we review and measure them. This page is the protocol those measurements follow, the rubric that turns them into scores, and the list of things we refuse to do.

One thing to know up front: the bench is being stood up now, and reports published before it produces data say so plainly. Until a report carries a benchmark table, its judgments are framework judgments built from platform documentation and the documented experience of operators, not measurements. We will never present a number we did not measure or cite.

The bench

Every host in a benchmark cycle gets the same treatment:

• Identical test sites. The same WordPress version, theme, plugin set, and seeded content on every host, on the plan tier a real small business would buy.
• Scripted checks, not hand timings. Time to first byte, full page load, load under concurrency, and cold start after idle, each run repeatedly from the same measurement locations and reported with the run count.
• Scheduled re-runs. Hosting performance drifts as vendors change infrastructure, so benchmarks re-run on a schedule and reports carry a visible updated date.
• Published conditions. Each benchmark table states the plan tier, region, and test dates it was measured under, so anyone can reproduce the setup.

The scoring rubric

Scores run 0 to 10 and weigh five dimensions. Measured speed carries the largest weight because it is the dimension we can verify directly:

• Measured speed. The bench results above, weighted toward performance under load rather than idle best cases.
• Reliability. Observed availability of our test sites over the measurement window. Vendor uptime figures are treated as SLA claims and attributed, not adopted.
• Real cost at renewal. What the plan costs after introductory pricing ends, with limits and overage behavior counted. We do not publish specific prices because they change too often; we describe the structure and link the vendor page.
• Support quality. Response and resolution experience on real tickets from our test accounts, plus corroborated operator reports.
• Platform limits. What the platform disallows or throttles: plugin restrictions, PHP worker ceilings, visit counting, staging behavior.

What we will never do

• Invent numbers. Every figure in a report is measured on our bench or cited to its source. If neither is possible, the report says the data does not exist yet.
• Let commissions set a score. Affiliate relationships are disclosed on every monetized page, and the rubric is the same for hosts that pay us nothing.
• Hide who we are. This site is operated by the maker of RankFlywheel and ProvenSEOTools, and pages that promote those properties label them as ours.
• Publish unreviewed machine output. Research and drafting are AI-assisted; every report is reviewed, fact-checked, and edited by the named human author before publication.

Updates and corrections

Hosts change infrastructure without announcement, so a benchmark is a dated snapshot, not a permanent truth. Reports carry the measurement window, re-runs replace stale tables, and if we got something wrong the correction is noted in the report rather than silently edited.