When evaluating a QB, I look at his skills and ability to make the plays necessary to accomplish tasks which lead to winning the game. This isn't an easy task because in order to do it properly, one has to examine only what the QB contributes to the play, and not what other players do which may make the play work (while that may show THEIR skill or ability, it doesn't mean it shows the QB's skill or ability).
To make an extreme example, if a QB chucks a horrible pass up for grabs and his receiver comes down with it and darts around five guys to score a TD, it does not mean the QB made a great play just because the receiver used his skills and a good team result occurred.
Some skills are easy to assess, such as foot speed and agility. Clearly, a player such as Michael Vick or RG III demonstrates they have these skills on a fairly regular basis by avoiding the rush and running for large yardage gains among other things.
Other skills are more difficult to figure out, such as passing accuracy under pressure. First you must determine what constitutes "under pressure", then you must assess passing accuracy by something more than whether or not the pass was completed (completion percentage), as this is a failed metric when it comes to measuring the skills of the QB (dropped passes lower it, shorter throws and receivers beating the defense and getting wide open increase it, and there are many more variables as well).
The only true way to assess a QB's skill is to examine the plays that work or do not work (to whatever extent) and figure out why the outcome ended up being what it was (why did ithe play succeed or fail to the extent it did). In short, one must determine how much the QBs actions on any given play affected the outcome of the play. Many people incorrectly assume since a QB touches the ball every play he must have the most impact on every play. While he does have the most overall impact, many times the QB is not the most impactful player on a given play.
So if you want to measure true passing accuracy under pressure, you must first identify situations as being "under pressure" where the QB throws the football. Second, you must determine the outcome of the play and how much the QBs throw contributed to that outcome (good or bad).
Finally you have to draw a conclusion. So, if you're evaluating great plays in an effort to determine QB greatness, you have to figure out as best you can the answer to this question: "How many other NFL QBs could have made that exact same play if they were in that exact same circumstance?" If the play seems great but a lot of other QBs could have done it the same way, then it's not anything special. However, if not many other QBs could have done what this one did, then it could indeed be a sign of greatness.
To make things simple, when I look at most QBs people consider to be great, it is fairly easy to find plays with an enormous positive outcome for the team where the QB's skills or abilities were critical to the play happening. There also tend to be a small number of plays which fail that are attributable to those same QBs.
However, when I look at Brady, while I DO see a small number of plays in which failures are attributable to him, I also note that his biggest failures 1) tend to come in bigger moments during the game and/or in big games; and 2) tend to be overlooked by many observers at a much higher rate than the same mistake would be overlooked for other QBs.
I also see a startlingly high number of plays which are successful where Brady's actions are so ordinary the vast majority of NFL QBs could have done the same things he did. That doesn't signify greatness; it merely shows he didn't screw things up, and hype assigns greatness where it doesn't actually exist.