I worked at a startup for a bit that started OKRs and I found experience super enjoyable; to the point where working in a more relaxed environment(for lack of a better description) is kinda hard these days.

The two things people and teams struggled the most with though, IMHO, is:

A.) Coming up with a good OKR that met longer term, higher level company priorities that intersected your work.

B.) Coming up with a good ORK in terms of measurable key results.

It can be tough and time consuming to get the measurements in place necessary to gauge the results, or even make a case for the OKR in the first place(incidentally this is the dirty secret of SRE IMHO). We would pro-actively get data collection in place on occasion for reasons that included helping to launch and score future OKRs.