It's terrible. Don't do it. I took an order for a large number of double sided campaign banners for a local election back in 2014. The concept of doing it is sound. You use test prints of a thin rectangle the same width as your banner to set the offsets in your RIP so it prints truly centered on the banner. You create a mark on the output heater where you will always set the vinyl to start a roll of prints (you cannot start the vinyl under the lid because the reverse side curls upwards and will catch on the lid). Print the run, then dry the print overnight, reroll, print the reverse side.
In practice, the problem is that the printed side of the material tends to go through the printer unevenly when flipped over. The stickiness of the printed vs. unprinted areas is different, so the media tends to bunch and wrinkle as it goes through the printer, causing headstrikes, then nozzle dropouts. The likelihood of the media picking up lint or fuzz during dry time is high, and that fuzz catches on the printheads and causes ink drips. It's easy to not roll the media back up perfectly straight and tight, leading to skewing.
We probably lost 20% of the banners we attempted to print, and took more than twice as long to complete the order as I had anticipated.
A lot of this may depend on the actual media you are using. I have in the past used some two-sided banner material that was stiffer than what was available for that particular order, and it was a little easier to use, but still prone to the same kind of problems.
Accuracy of print location isn't too bad. Most times we were within 1/4" to 1/2", which works out fine.
