01
Step one: fix the conditions before you start
Choose three to five pages that represent real use, such as a news article, a shopping page and a media-heavy homepage. Use the same device, browser, network and time window for every run. Clear the cache before each cold load, and note whether you are testing cold or repeat visits. Close other heavy applications so they do not distort timing. Write the setup down in one short note so someone else could repeat it. Worked example: a freelancer tests four pages on one laptop over home wifi, each loaded five times with the tool off, then five times with it on.
02
Step two: measure the same thing every time
Pick one or two measures and stick with them. Browser developer tools show total load time and when the main content appears, and those are enough for most comparisons. Record each run in a simple table with columns for page, tool state and time. Discard nothing quietly; if one run is wildly slower, mark it and keep it visible. Using the median of several runs reduces the pull of a single odd result. In the freelancer example, the median of five loads per page becomes the figure for that page, which gives eight medians across the full test.
03
Step three: read the gap and decide
Compare each page's median with the tool off against the tool on, and look at whether the difference is consistent across pages rather than dramatic on one. Consider what you gave up: some speed settings block scripts, images or tracking that certain pages need. Open each page and check that it still works. In the example, the freelancer finds faster loads on three pages but a broken checkout on one, so they keep the tool for reading and turn it off for purchases. Repeat the test after updates, since results change with browsers, networks and site design.