Neither strategy is sufficient. Putting armor on areas that had statistically more bullet holes based on examining planes that survived, is obviously not a good strategy as explained in the article. But neither is the strategy of armoring up all the areas that did not have bullet holes. May be not all of those areas are critical. You just don't know unless you examine the planes that did not survive.
In the absence of data on the planes that did not survive, you just have to follow sound structural/mechanical/electrical/mathematical (flight trajectory) principles to determine which of the exposed surface areas of the plane are likely to be hit under various trajectory fly overs over anti-aircraft gun batteries, and among those areas , which are critical (engine, pilot cockpit, weapon storage, fuel etc), and armor those areas up.
In the absence of data on the planes that did not survive, you just have to follow sound structural/mechanical/electrical/mathematical (flight trajectory) principles to determine which of the exposed surface areas of the plane are likely to be hit under various trajectory fly overs over anti-aircraft gun batteries, and among those areas , which are critical (engine, pilot cockpit, weapon storage, fuel etc), and armor those areas up.