Web performance has a good case of *it works on my machineitis*. This doesn’t always happen out of wrongful intentions. It feels good to share a high Lighthouse Performance Score or a test that passes good benchmarks! But there’s a fair bit of nuance here.

If we’re privileged enough to own high-end devices, our local tests are likely to be favourable. If we monitor only on iPhones and fast, desktop cable connections, we’re likely to see a more positive outcome too. Our choices (and contexts) can easily hide poor experiences for large portions of the audience.

To improve speed across for everyone, we need to test comprehensively, preferably in more adverse conditions. Also, we need to remember that [two tests in different settings](https://calibreapp.com/blog/common-mistakes-in-tracking-speed#expecting-the-same-results-from-different-tools) are far from equal. **Where and how you test matters. **