Attribution Analysis

Lately, there has been a lot of talk among marketing community about attribution analysis. Put simply, attribution analysis, is a method of allocating business goals to specific marketing tactics, i.e. TV campaign, website design, SEM, etc. In theory, this is a very useful tool, as it would allow marketers and advertisers to finally pound their chest and say "I told you so! What we are doing is relevant to your bottom line!" In practice, however, it is not that simple. Designers don't design for things to be easily measured, or, evenmore, customers are not easily forthcoming in their cross-channel behaviors.

I considered doing attribution analysis once, when I began my journey into Marketing ROI measurement. I decided to start with the (in)famous purchase cycle funnel (or hourglass depending on whom you ask) and use conversion rates between each step to calculate the attribution rates to each step. Then, as theory went, I would break out each channel within the step based on individual conversion rates, and voila -- I have my attribution model. Sounds good. In practice it was a bit more complicated as one can imagine. While I was working on perfecting my thoughts, I met Riccardo Zane, who gave me a new view on the purchase funnel and asked me to do what I've always been preaching -- look at it from consumer's point of view.

When we illustrate 'customer journey' through the funnel, we never illustrate it linear, it always jumps from one place to another. Yet, we never ask, why is that? If that is the case, maybe the funnel (hourglass) is not the right model? Maybe, customers don't make their decisions the way that we always have modeled? And if that's the case, what about attribution? Does it make sense at all?

Of course, people assign different channels different values when they make their purchase decisions. It is obvious, even by looking at such unscientific methods as ourselvles. From the marketing intelligence standpoint, do we care? Yes, one can argue, that if I know which channel is most important for my customer to make a decisions, I can invest more of my budget into that channel so that more customers can make a positive decision and make a purchase. I would think, on the other hand, that what we care more about, is knowing what actually makes a person decide to purchase, what kind of intelligence they are looking for.

Let's remember, that as marketers we posess a lot of channels of information, and very few channels of actual purchase. Before late 90's we had 3: in-store, mail-order, and phone. Then we've added web, and to the lesser part (at least in US) mobile. Web added an enormous weight to marketing efforts, as it required both the need to provide real-time information and real-time purchasing ability. It allowed customers make spur of the moment purchases (probably contributing somewhat to the current economic environment), but also, added to the information overload. How do we account for all this?

I say we don't. Let's take the simple approach. What do we care about? Customers buying our product repeatedly, our brand to increase in value, and more customers joining our fold. And we want to do it as efficiently as possible. The key to this is -- Customer Intelligence, which is to say understanding what our existing and future customers need to make a decision, and what our non-customers need to change their stance. It's not about attributing 10 purchases to facebook page, or 100 sales to a TV campaign, it's about understanding what customers need from across channels and providing them a unified, cohesive message that would allow them to make a positive decision towards our product again and again.

I would argue, that from optimization standpoint, we should stop attributing KPI's to each channel and start measuring and optimizing channel cohesiveness, campaign impact and across channel immersion. How do we do it? Stay tuned...