One possibility that’s disturbing for people who work in the media but probably worthy of being taken more seriously is that the reason ad rates for web ads are so low is that advertising is actually much less effective than people have historically thought, and the greater measurability associated with web advertising is just revealing that fact. Check out, for example, this Ecocomics discussion of Spider-Man’s short-lived efforts to advertise on his own behalf:
However, Spider-Man shouldn’t fret too much. In actuality, it is not likely that advertising and branding would have really altered public perception of the web-crawler. A recent study (not sure if there is an ungated version) by Robert C. Clark of Harvard University examined a panel data set of advertising expenditures for over 300 brands in order to determine the effect on both brand awareness and perceived quality. The overall results suggest that advertising expenditures have a significant effect on awareness but no significant effect on perceived quality.
There does, however, seem to be a noted distinction once measuring the effects for different categories of products. For example, expenditures for the fast food industry had a marginal effect of 0.0144 on brand awareness and a marginal effect of 0.000727 on perceived quality (which is pretty large compared to the rest of the categories).
There are some markets where upping awareness while not impacting perceived quality can be very useful. This is, I think, why car insurance constitutes such a large proportion of TV advertising. But in general, it strikes me as telling that ad agencies have historically invested a lot of energy in “wooing” clients and relatively little in developing useful metrics of their own success.