20V absolute maximum mean that is the voltage at which they guarantee every part will still function. Generally most will survive slightly higher voltages, at ambient temperatures, but over the full operating temperature range the 20V is what all the devices will survive at least for 1000 hours. you can get away with it for your particular samples, but another batch might not be so lucky.
I had a computer where the 5V power supply decided to go out of regulation, applying the unregulated 14V input ( the designers probably wanted a lot of margin in the supply, and it certainly would survive a really big input Ac voltage variation of -50% to 100% of the rated 115VAC 400Hz 3 phase input no problem short term) to the entire board of TTL and NMOS IC's. Only failure was a single hex TTL 30V rated inverter, which did not like having 14V applied to it and went first short circuit, melted the flatpack lid off and then blew it's bonding wires off. The rest of the TTL all survived, and in fact were still operating correctly logically, and still gave valid results. Change the board ( lot of them spare, they rarely gave issues so plenty of new old stock around) and tested after replacing every part in the 5V power supply, including the failed crowbar that should have blown the fuse when the voltage went over 7V. Was not going to look for the failed part, easier to replace every part, including the resistors and the ceramic capacitors instead. only things not changed were the 2 unobtanium block tantalum capacitors in there, 8200uF 25V and 10000 uF 16V, as they were a lovely Vishay part, and lead times on those were in the decade range, seeing as I had orders for that long when I arrived there, and they never turned up. Saw them in the online Vishay catalogue years later as well, with a "contact factory for price and lead time" as the only ordering information. the others were replaced with regular 105C electrolytics, as they would last as long as the rubber seal wet tantalums in service.