DebugBear tests page using Lighthouse, as well as running custom logic.
There are a bunch of differences between Lighthouse and Lighthouse on web.dev, PageSpeed Insights, or Chrome DevTools:
One keys difference is that DebugBear directly throttles the computer's network connection, while Lighthouse normally loads the page without throttling and then simulates how it would have loaded on a slower connection. I wrote a blog post where you can learn more about the different throttling methods.
JavaScript throttling by Lighthouse is also normally simulated, while DebugBear use Chrome to slow down execution.
DebugBear's throttling is more realistic and gives you more opportunity to debug your problems.
The performance scores by Lighthouse gives you a better absolute indication of whether your site is fast. The scores measured with simulated throttling are calibrated against average performance across the web. If your site only gets a score of 65 on DebugBear but a score of 90 on web.dev that means your site is pretty fast.
Use DebugBear to track performance, identify opportunities, and understand why regressions happened.
If non-trivial bits of JavaScript code are running regularly after the initial page load you may get an NO_TTI_CPU_IDLE_PERIOD error from Lighthouse. Lighthouse is waiting for the CPU and network to be quiet, but never reaches a point where it considers the page fully interactive.
For example, this might happen if you have an JavaScript-based animation on your page.
To solve this issue, either remove or disable the code that his occupying the CPU, or disable it, for example by adding a query string like "?noAnimation" that is then acted on by your application code.
The "non-trivial" threshold used by Lighthouse is 50ms of execution time.