Project Wonderful is probably many people's first experience with placing a display ad. But as far as I can tell, exploration of the science of advertising by webcomics people has been pretty superficial. That includes me, so I decided to do some research.
I didn't crack open any marketing texts, because I don't have any. Most of what I've read about economics comes from journalism sources, so I turned to places that report on scholarly papers to see what is being written.
My first surprise was the assertion that ads not only stop working, but eventually begin to backfire. The idea is that new ads draw attention because they appear scarce and thus more valuable. Eventually, when they become familiar, this value is lost. When they become sufficiently commonplace, they are perceived as having little value until eventually they come to be something to be pushed away.
This suggests why we have the concept of advertising campaigns, as tired promotions lose power and must be replaced.
This phenomenon is described in a bell curve. I've seen my own successful ads stop performing well over time, and I generally give the site a breathing spell or change the ad. Some sites, however, have a lot of new traffic coming in, muffling the decline and making it harder to spot. The ad might continue to perform adequately from newcomer traffic, but it may possibly already be repelling people who are sick of it. though it may be difficult to rotate or idle a performing ad, it may be a good idea.
Another line of study tried to update the notion that people are passive observers of ads. This strikes many as an obsolete idea, and indeed, it seems the opposite is true. Eye motion studies suggest that a person's sensory and cognitive interaction with an ad may be directly related to that person's goals. Those goals are extremely varied, embracing almost any agenda.
A webcomic ad viewer may have a goal of spotting new comics that will be of interest. An ad that telegraphs quality in art, design and message may receive a click.
Another viewer may be looking for horror comics, or may be provoked by ads that cause curiosity. We can't guess the agendas of all viewers, but we can produce ads that are a fair representation of what lies beyond the click. A clever ad may bring a surge of traffic, but so will placing dozens of ads. Off-message ads may be a waste of money compared to honestly representative ads that attractive fewer, more loyal, visitors.