Some of it is the level of study. For example, only a few studies are cited in the TimeTree for bivalves. But studies on bivalves that are cited for some pairs are not cited for other pairs covered in that study - there seems to be some lack of knowledge about the groups, leading to inaccurate integration of data into their database.
But if you pick a more frequently studied pair, say Homo and Pan or Homo and Mus or Homo and Didelphis, and look at the range of dates given for the divergence by individual studies, it does not inspire confidence that molecular clocks are reliable. There is a huge amount of data behind it, but little evidence that the algorithms are actually reliable and inadequate effort given to data verification for the calibration points.
Also, there are some common sense checks that are missing. Larger clades are not necessarily reported as older than their subclades, for example.
Likewise, the OneZoom tree no doubt is reflecting data deficiencies in some of its strange groups within Mollusca. But it has some claimed groupings that are wrong, rather than merely a jumble of unknowns. I haven’t had time to dig and figure out where the problems are that lead to putting a pulmonate land snail in Vetigastropoda (on time divergence, about like putting a mammal in as a shark). But there is conspicuously no one involved that knows much about mollusks.
The difficulty is that the presentations claim to be authoritative without taking responsibility for data quality. If I quit my job and quit doing my research, I could contribute a lot of corrections to such sites, but there are some practical difficulties relating to staying fed and having access to the literature with no income.