WA/WB is often synonymous with pot control.

The basic characteristics of a WA/WB situation are these:
1) You don't know if you're ahead or behind
2) If you're behind you are unlikely to improve so you don't want to put more money in the pot
3) If you are ahead you are unlikely to get called by worse if you show aggression

It's interesting to consider betting for information here. Let's say we bet for information.

If we're WA we learn that we are WA because our opponent folds. The value of this information is in that it tells us that we had the best hand and we failed to extract any value from it. We learn that we were wrong to bet.

If we're WB we learn that we are WB because our opponent calls or raises. The value of this information is in that it tells us that we had the worst hand and we invested more money on the worst hand. We learn that we were wrong to bet.

WA/WB situations are classic examples of the problem with betting for information.

The more the hand is checked before we act, the more likely that our hand is the best as every check narrows our opponents range a little bit in the direction of weaker/missed hands.

Similarly, the more we check, the weaker our perceived range is. This means the more we check, the wider our opponents calling range becomes with worse. Likewise, the more we check the wider the opponents betting range becomes with worse.

As Stacks said so well, like basically every poker decision the most important thing here is a range consideration. What range did villain get here with, and what part of his range will he take which actions with. If we know this, we can calculate exactly how many bets it is optimal to make or call, and if it's optimal to make them or call them. As a rule of thumb though, the number will tend to be between one and two.

Stacks also talked about range balancing with checking AK/AQ occasionally. Another thought that occurs to me is range balancing by betting KK and checking A-rag. I haven't really thought this through yet though and it may be completely misapplying the range balancing concept.