“Rather than two philosophical combatants misunderstanding and mistrusting each other, I see [marketing mix and attribution modelers] sharing insights and learning from each other.”
– Jim Sterne, “A Letter from the Chair…”
As with all things, new technology gives rise to new data and thus new modes of analysis. A new ecosystem forms. Marketing mix models were necessary when data only came aggregated. When tv commercials were one-way conversations and data vendors recorded their performance regionally. Retailers aggregated purchases at the store-level, and customer loyalty behavior wasn’t stored on a database for longitudinal, individual-level tracking. Consequently, this aggregated data made it possible to do inter-channel (“horizontal”) marketing spend analysis. Marketing mix was the name of the game, and it was prime for the technology available.
Attribution currently applies to a stream of exposures and clicks, and attributing the value of each to an observable outcome. This is excellent for single-channel (“vertical”) decision making, but we are just beginning to find the right places to apply attribution models. As the technology for two-way dialogue between brand and consumer proliferates, the data stream will not only consist of broader online behavior (exposures and views; social and direct), but also tv ad exposures, location-based ad exposures (e.g. the opportunity to view a billboard on a commute to work via location data from a mobile device), and telephone interactions (the major telcos now provide phone, tv and internet service). At the very least, granular multi-channel data will be available at the household-level. At the inevitable best, at the individual-level. It’s only a matter of time.
That said, we are neither here nor there. We sit at the intersection between aggregate and granular data, horizontal and vertical analysis, and marketing mix and attribution modeling; by virtue of the technology available. So while the two schools of thinking can learn from each other, one is evolving from the new data-technology ecosystem while the other is attempting to adapt to accommodate it.