• by August 15, 2008

Excellent piece [from Josh Chasin]. You've asked a question that folks in media have been asking for as long as I have been working in research - and I was around when computers had their own plumbing.

Back in the '60s, there was a fair amount of research done on "how advertising works." There were a number of well-thought models that came out of the effort - the one I remember most was called DEMON (for DEcision Mapping via Optimum Networks). These models all had one flaw: they relied upon a "magic" ingredient, called the "util." This was the weight of the effectiveness of other factors to the consumer, and like alchemist's gold, was never really defined or produced.

The '80s brought PIMS (Profit Impact of Market Strategy) a real attempt at quantifying what works (or doesn't) in a marketing effort. PIMS followed the varying success of 3000 business units in 450 companies, and attempted to map best practices to success in marketing. The book on this study (by Buzzell and Gale) is still good reading.

Since then, there have been a number of runs at mapping the advertising mechanism, ranging from behavioral/psychological models to higher math. None have really given good, useful, work day answers to the question.Media market modeling, where the most granular ad expenditures are recorded and compared by brute computer force seems to shed some daylight on what is happening - from one point of view, anyhow - but only the largest businesses can afford it. In any case, those of us in media have little hope of getting our advertisers to show us what's going on at that level of detail.

Most recently, the folks at ARF - frantic at the decline of the 30 second spot and other high value legacy marketing tools - came up with the idea of an "engagement" factor. "Engagement" can be used to explain why one ad campaign works while another does not. In the end, it sounds a lot like the old "util" to me.

I think [Chasin is] right. Each of us will eventually develop our own take on what works ... kind of a private PIMS, if you will. As long as the tools are good enough, we'll be able to see what's working and what's not. The key will be to know when what used to work is no longer any good.