• by Josh Chasin , Op-Ed Contributor, August 5, 2008

I've been in a lot of conversations and meetings recently where the subject has turned to the question of "How does advertising work?"  Invariably, when this topic comes up, you'll hear two things: someone providing empirical quantification of the ROI of a particular type of advertising (or of an entire campaign); and, someone bemoaning the fact that, at this late stage in the evolution of marketing science, we still don't really know -- or can't prove -- how advertising works.  

We've come to accept that the Internet is the "most measurable medium," largely because if you expose a consumer to an ad and invite him or her to click on it to come buy something, then we can measure how many times that ad was served and how many things were sold on the click-through.  

This troubles me.

It troubles me because all of online advertising seems to end up being, either overtly or tacitly, about the click-through. It's this kind of logic that would hold that Yellow Pages advertising works better than network TV advertising, because the Yellow Pages makes the phone ring (the analog ad men called that the "call-through.")  And yes, Yellow Pages advertising works great -- but if the Yellow Pages are the offline search advertising, then offline display attracts many times as many ad dollars as offline search, which I think suggests that there is an awful lot of upside in online display (including video and other rich formats)... providing that we can convince ourselves, in this clicked-out world, that such advertising actually works.

Lately in the online advertising space there has been some great work done around what Microsoft/Atlas calls Engagement Mapping, and what has been referred to as the assist model, or going beyond the last click, or attribution modeling -- allocating credit for a click-through to all the touch points that preceded the one that apparently closed the deal.  This is great stuff, and helps advertisers to get at the holistic impact of their campaigns, and to understand how impressions that don't immediately trigger a conversion still contribute value.

Still, I feel like we are collectively discounting the value of ad exposures that occur further out from that conversion (and too, there's a whole ‘nuther bucket of online exposures that precede an offline purchase, and I don't know how we develop a system to routinely, holistically credit these impressions.)  Consider, for example, the category of automotive.  Suppose the consumer buys a new car every three years, and is "in-market" the last six months before the purchase.  Do the two-and-a-half years of car ads that the consumer has seen prior to the "in-market" phase have any value whatsoever in the ultimate "conversion"?  I would argue that they most certainly do; the consumer's consideration set is formed over time, especially for such a major purchase, and the cumulative effect of those exposures has an impact on the consumer's attitude toward the different brands.  

So how do you account for the value of those impressions?  How, exactly, are they affecting the consumer's cognitive processes over time?  The way the consumer thinks about, feels about your brand... but that squishy, touchy-feely stuff ends up being harder and harder to quantify in terms of ROI, and I fear it often gets lost in the shuffle... even though it remains profoundly important.  Perhaps, in this cluttered, over-communicated age, more important than ever.

When I was in college studying marketing, one of my professors told us that "it is impossible to measure the effectiveness of advertising."  Impossible?  I was incredulous.  I started to argue, but then I remembered, "Hey, it's only 1980, no one's even heard of single source yet."  So I just asked if it would be on the final (it was.)

Buying a car is a major, well-considered, big ticket purchase; let's think about the CPG category.  I do the grocery shopping in my family, and I have a four-year-old daughter, so I buy a lot of  Kraft Macaroni and Cheese (the classic, what they call "blue box") because my kid lives on the stuff.  I buy it on my Sunday trip to the supermarket, provided we have run out.  As my friend Erwin Ephron says, the purchase cycle for cereal is, the box is empty.  (Bookmark  Erwin's site; you'll thank me.)  So does any advertising contribute to my mac'n'cheese "conversion"?  I would argue, probably not; my purchase is driven entirely by purchase cycle.  And yet, I am absolutely certain that Kraft has reams of data that quantifies beyond any reasonable doubt that weekly sales are highly correlated to advertising.

So how does advertising work?  

I've come to think that the question itself is a false premise, because there is no one way advertising works; there are a myriad of ways.  Are you trying to get heavy category buyers to switch to your brand?  Do you want existing brand buyers to buy more often?  Are you selling on price?  Luxury?  Service?  Are you in a highly commoditized category, or a heavily differentiated one?  How long is the purchase cycle?  Do you only want to reach consumers in the purchase cycle, or are you engaged in long-term brand building?  All these questions and more will impact the way your advertising will work (and I haven't even mentioned engagement...)

So it is no wonder that we can't answer the deceptively simple question, "How does advertising work?"  There is no simple answer, and that should be neither surprising nor daunting.  It's like Curly's "secret of life" in "City Slickers."  How does advertising work?  That's for you to figure out.  And it's not like there's any shortage of tools.

How does advertising work?  Hey -- you tell me.