Ah well. Here is some of the argument on blur in another context. I've looked at it a few times since, and it is very efficient where a character has good AC to begin with. D&D 5E - Bladesinger - a criticism of its design [EDIT let's not get into bladesingers though! This is linked only to share with you where some previous discussion landed.]A very specific situation can be analyzed with probability distributions (including crits). However, the analysis for this could change significantly depending on the specific damage, hit rate and number of attacks paramaters you use. IMO, that means that specific kind of analysis in regards to this question won't be generalized enough to be beneficial.
Another way to model this is to consider the reduction to the number of hits you are taking. If your AC is such that monsters hit you on say 18+, then gaining +1 means that for every three hits you were going to take, you instead take two. An improvement of a third or 33%. Generally, it is more important to model defenses as reduction in incoming, than as change on base value, because it is the reduction on incoming that will be experienced at the table.This is only true if you are talking relatively. Absolute matters more given the number of hits we are talking about potentially receiving. +1 AC will cause the same number of hits to miss at 20AC as at 10AC.
One character in my current campaign is an SnB battlemaster. The player isn't overly concerned with optimisation, but they did take commander's strike, which they can use to give the party rogue a second sneak attack each round. The battlemaster simply stands next to a target, and it will be in for two turns of sneak attack damage a round (presently +6d6 per sneak attack). The battlemaster has chainmail, a shield +1, and defense, for 20 AC. The character's weakness - as a dwarf - is their move.
In a recent encounter - in ToA - the monsters had +6 to hit, and there were nine of them with two attacks each. For every eight hits he would take at AC 19, he takes seven at AC 20. An improvement of 12.5%. Their damage was about 10 a hit, and due to specifics of ToA he has 144hp. On average it would take 41 of their attacks for them to kill him at AC 20, and 36 at AC 19. Again around a 12% improvement. More tellingly, if they dog-piled him they will on average kill him in two rounds, at AC 19, but will need three rounds at AC 20. If he had plate, the benefits of defense would be even more pronounced (about 17%, instead).
These kinds of considerations are not obvious, and I guess bring me back to my theme that defenses are typically undervalued.