some good stuff

There are a couple of excellent discussions going on right now. Jason has put up a series of great posts about pre-testing that have even engaged Nigel Hollis at Millward Brown.

I've probably been through more pre-testing this year than I care to think about and I've found that getting into arguments about whether it predicts in market results or not are slightly besides the point. Predicting in market results is virtually impossible, whether it's with pre-testing, econometrics or voodoo. Lets also put aside the issue of whether these tests use a model of advertising that reflects reality given that this is something in flux (or that we don;t know).

Barring a real in market test (which I've gotten some of my clients to do), the best thing a pre-test might ever do is teach us how to make a spot better. After all, isn't that the role of research - to make us smarter. This is where the quant tests fall down for me. Even if I believed that they worked (I don't) they do not give us any clue about why a spot is not connecting with people. Granted, finding that out in qual means some careful probing and knowing what to listen (and not to listen) to - but at least there's a chance of doing it.