I haven't used Dexcom, only libre.
Subject to that caveat, insertion causes some local damage, as you would expect with a needle being poked in, and the defence/repair mechanisms influence readings a lot till the repair etc. calms down: the cells involved eat glucose big style so it looks like bg is low in the sensor area even though it's higher elsewhere.
It's trial and error, really. I get pretty good results if I insert about 12 hours before activation, others have reported 24, 48 hours. A couple of hours seems too small, there'll still be a lot of repair work going on then.
Activating when stable seems to make a difference, makes sense seeing as stability is the time when bg and ifg are likely to be closest. Maybe not an issue with dexcom as you can calibrate it as and when you like (can't calibrate a libre, except in my head) although my recollection is that dexcom should be calibrated when stable.
There's an interesting paper on the complexities of it here:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2903977/
Re your avatar pic, you're really going to have to write a paper on the performance of
in vivo glucose sensing while hanging upside down. It really hasn't been studied in depth! Make it your field!