This is a great question, I have asked a similar question before, but got no answers. With that said, that was a more philosophical question and yours is more to the point.
However: I have a solution that have been tested on multiple occasions to IEC 61000-6-2
(note; IEC 61000-4-2 is just the method; the actual requirements would be in a different standard)
The simple thing to do would be to protect the TVS diode, typically with a polyfuse, so that it can survive a DC overvoltage, but that defeats the entire purpose of the TVS: Any component in series with the TVS is going to face the full fury of the surge pulse. I have not found any polyfuse rated for surge currents of several hundred amperes.
My solution has therefore been to rate the TVS so that a destructive DC overvoltage is unlikely. In your case that would perhaps be 15V. Then you need to have a secondary overvoltage protection. You can do as you describe and use a zener and the previously mentioned polyfure, but I find it more elegant to use an overvoltage lockout. As you are working on 5V you have lots of options available in the form of Efuses.
With all that said I have claimed before on this site that a polyfuse is quite common to place infront of the TVS, (That claim was shot down the last time, yet it is the first comment here) so I am curious to get more input on this.
simulate this circuit – Schematic created using CircuitLab