How we measure
Every number on Benchnode comes from our own lab. This page explains how we measure, so you can put the results into context and check them yourself.
The lab
Our lab is a separate WordPress install on shared hosting at All-Inkl in Germany. That’s the kind of hosting many of our readers use, and the same kind this blog runs on. The lab is locked from the outside and not indexed by search engines.
The reference page
Every test runs on the same page: a typical Elementor home page with a large hero image, testimonial slider, gallery, FAQ, contact form and popup, plus a typical blog post. That way we see how a plugin affects a real page, not an empty one.
The process
- The lab is reset to a saved base state. Database and files are exactly as before, and leftovers from earlier tests are gone.
- We measure the baseline.
- We install the plugin and turn settings on one by one. After every step we measure and check that menu, slider, popup and form still work.
- At the end the lab is reset again.
How we measure
We measure with Lighthouse, Google’s measuring tool that also powers PageSpeed Insights. Once with a throttled mobile connection, once on desktop. Each measurement is five runs, and we publish the median. Before the runs we load the page once so caches are filled. Tool, version, date and time are listed with every result.
Connections to the outside
Our own plugin Connection Monitor always runs in the lab. It records which plugins talk to outside servers and what they send. So for every test we can also tell you where data goes.
The metrics
- TTFB: time until the server sends its first response.
- LCP: time until the largest visible element has loaded, usually the hero image.
- CLS: how much the layout jumps while loading. LCP and CLS are part of the Core Web Vitals Google uses to rate user experience.
- TBT: how long scripts block the page. In the lab it’s the best indicator of how fast the page reacts to clicks (INP).
- Page weight and requests: how much has to be loaded.
Limits
The lab and the blog share a server with other customers of the host, so individual runs vary. The median of several runs evens that out, but not completely. Our numbers compare well within one test. Compared with measurements of other sites or tools, they are only partly comparable.
Independence
We are funded through affiliate links. That never affects a measurement or a recommendation, and we don’t sell placements. More in our affiliate disclosure.