How would this metric work for websites that load real-time dynamic content? For example, if a website loads quickly then slowly streams data from the back end, how would this metric account for that?
I think at some point in time they assume that the page is complete and use that as a reference point to compare the rest of the frames to. It's most likely the window load event.
I think in fully dynamic pages, that constantly change, it's worth presenting the video and let the person running the test pick a "final state" to compare against. You can then work backwards as in the automatic case.
I think in fully dynamic pages, that constantly change, it's worth presenting the video and let the person running the test pick a "final state" to compare against. You can then work backwards as in the automatic case.