I'd be happy with anything within like 0.04-ish. I doubt they're that accurate anyway. User-error in itself is probably 0.1x -- i.e., you test it, I test it, both with the same gauge, but still somehow come out with slightly different numbers. Carbon dioxide is a wild-card as well, it's constantly dissolving itself in there so the numbers will tend to bounce a little bit no matter how careful and consistent you try to be.
And then, of course, ask yourself... does it REALLY matter if your pH is 5.23 or 5.27?? Doubtful!!