Trust layer
How we test: one store, every host, no shortcuts
Anyone can rank hosts by reading sales pages. We rank them by running a store on them. This page is our whole method, in the open, so you can judge the numbers before you trust them.
The test store
We built one WooCommerce store and froze it. Every host gets the exact same copy, restored from the same backup, so the only thing that differs is the host.
- WordPress and WooCommerce, latest stable versions, recorded per run.
- The Storefront theme, unmodified. It is WooCommerce's own reference theme.
- A fixed plugin set: WooCommerce, WooPayments in test mode, Yoast SEO, WPForms Lite, UpdraftPlus. Diagnostic tools come off before we measure.
- 100 products in 10 categories, with the same product images at the same sizes on every host.
- Five pages under test: home, shop, one product page, cart, and checkout.
What we time, and why the admin is on the list
Most labs time the homepage and stop. Store owners told researchers, in thread after thread, that their daily pain is the back office: an orders page that takes six to eight seconds, a product edit screen that crawls once a store holds thousands of orders. So our protocol times three surfaces, not one.
- Storefront speed. Time to first byte and full page load for home, shop, and product pages. Median of repeated runs, one fixed test location (Dulles, Virginia), same browser and connection profile every time.
- Checkout flow. A scripted run: add to cart, open the cart, load checkout, submit a test payment. Timed per step, three runs. Checkout pages cannot be served from cache, so this is the honest speed test for a store.
- Admin speed. The orders page and the product edit screen, timed on a store loaded with order history. This is the test we added because owners asked for it, and the one most review sites skip.
- Load test. Fake shoppers ramp from 10 to 50 at the same time, browsing shop, product, and cart pages for 10 minutes. We record the error rate, the slow tail, and whether the host throttles or starts refusing shoppers.
- Uptime. A check every minute for at least 14 days per host. Every outage gets a timestamp.
- Support. One identical ticket to every host, filed at the same time of day, asking about checkout errors under load and PHP worker limits. We record the first reply time, the time to a real answer, and whether the answer was any good.
Our publication rules
- We publish only what we measured. A result that was not measured does not get printed, estimated, or rounded into existence.
- Every number carries the host, the plan, the date range, the tool, and the test location.
- Raw reports and screenshots are archived before anything is written up, and results link to them.
- Until a host's full run is done, its status reads "testing in progress." No exceptions, no previews dressed as results.
- We re-run the full protocol every quarter, and we re-measure a host that changes its plans or its platform instead of rewording old copy.
- Hosting is one of three causes of a slow store. Theme and plugins are the other two. Our results say what the host did. They do not promise what your store will do.
What the lab costs, and who pays
We buy the entry plan at each host ourselves, at the normal monthly price, the same way you would. Hosts do not get free placements, early looks, or a vote on results. This site earns money from affiliate links, explained in full on our disclosure page. If a host ever offers us a free account, the results page will say so next to its numbers.