# The Attribution Riddle: Work-In-Progress
‘I think we have a problem.’, he said. ‘No!’, I retorted. ‘This is only part of the problem.’
Just because we now have an attribution model and behavioral model doesn’t mean we have cracked the attribution riddle. Attribution is arguably the holy grail for all marketeers, and this problem gets even more compounded in a B2B or business-to-business scenario. All of us remember reading John Wanamaker’s famous quote from eighty years back, “Half my advertising is wasted, I don’t know which half.” The trouble became more acute in recent times as most marketing budgets got spread across the more traditional advertising and digital advertising. I could be spending millions on native advertising and another pot on outdoor but how am I to gauge the moment of truth in my buyer’s journey. That moment where she became favorably predisposed to my brand. Was it while walking to work from the train station where she glanced upon the wall on her right adorned by my million-bucks creative or was it when my inconspicuous display ad morphed into a piece of content appeared on her handheld screen while browsing her favorite news site? Or was it that both made a difference in subtle inexplicable ways and I’m only rationalizing the spend across the two post facto based on my intuition and where I want to go next with my budget!
Last mile models or ‘last click attribution’ doesn’t tell the whole story, for it veils the progression of a buyer along the journey up until the point of purchase. Applying data science and complex regression analyses may appear to be a smart way to distribute that sought after credit across your campaigns and channels. But it has it’s limitations especially when attempting to look across both of offline and online channels. And it gets even more convoluted when the points of purchase are both online and offline. But having a dynamic attribution model is quintessential to calculating your Return on Marketing Investment or ROMI.
As long as we recognize that no model is robust enough to withstand time lapses and long sales cycles especially in the B2B context. And that both the model and the underlying assumptions need to be constantly tweaked for us to see marketing performance across channels and over the lifespans of our campaigns.
I sense that Google will at some point let AlphaGo take a stab at solving problems which have a direct impact on Google’s business. A predictive marketing attribution model hat encompasses everything non-Google too will be welcome!