Georgia Pacific’s In-Sourcing Business Model is a Supplier’s Roadmap
Georgia Pacific, one of the nation’s largest paper products CPG, found a series of opportunities to improve their analytics processes and deliverables. The solutions that they came up with solved, at least in part, many of the intractable issues that they had just come to expect because of their reliance on data and analytics suppliers: cost, speed, and trust.
Their approach is a lesson for data and analytics providers to take to heart; GP’s approach can be applied by providers like DDG to help our clients scale their analytics operations, save money, and add tremendous business value to their analytics.
SPEED: There’s something inherently odd about a supplier’s taking 18 months to build a multi-touch attribution model; while it may be a work of analytical elegance, it’s essentially useless as a tool to direct marketing dollars because the data. Is. Just. So. Stale. As data and analytics providers, we should invest the time once: to develop scalable, automated, and meaningful analytics platforms for our clients’ use. The speed-to-market requirement may vary based on the application, ranging from several days to near- or real-time. In no case should it take 18 months.
COST: One of the biggest drivers of cost is the historical lack of the economy associated with scalability. For example, the prevailing thought was “how could an analytics solution that works for one brand possibly work for another,” even if that other brand is housed in the same building? The traditional answer to that questions was simply “it can’t.” That unimaginative answer enables vendors to create, from scratch, a brand-new solution (pun intended) at twice the cost. That’s great for the vendor but for the brand? Not so much. The challenge we suppliers have to face is convincing Brand A, who has acute interest in a data solution, is beneficial to the sister Brands B, C, and D as well. Then it’s contingent on us to make it scalable across the enterprise.
TRUST: Frequently, management severely penalizes data solutions from suppliers because they don’t understand how the work was done. This is particularly true in the world of advanced predictive modeling, where many data scientists like to impress their clients with their esoteric vocabularies, using words like “heteroskedasticity,” and “multicollinearity.” That’s actually less of a problem than the black-box nature of many of their modeling solutions. Data suppliers should take pains to describe our methods in business terms that mean something to our clients, not to ourselves. The ability to describe our models gives us the ability to defend our models—in terms that mean something to our clients’ business. We can break down that trust barrier by using solid data science, transparency, and great storytelling.