Data Science and Analytics has for long been harnessed in silos, both functional (marketing, finance, supply chain) and technical (databases, analytical software, Excel, etc). As organizations look to implement organization-wide analytics strategies, it is important that DS&A also moves out of the modeling and algorithm world.
As has been wisely said “Fall in love with the problem and not your solution”, so also for companywide adoption, experts have to think beyond “p scores” and “correlations”. DS&A has to be viewed in the context of an end to end business process. DS&A solutions are meant to solve business problems, hence the analyst must seek to understand both the business problem and the technology ecosystem that supports the decision making process. Keep the business problem in mind when building the solution. Yes, the data discovery process leads to some really good nuggets of insights and information, but it is important to be focused on the business problem statement.
DS&A experts need to understand the end-to-end technology architecture that enables analytics deployment. Deploying recommendation models based on customer behavior across channels is really dependent on recognizing your customer across channels, which is a function of robust MDM deployment.
Further, an omni-channel recommendation system needs integration between multiple business systems (call center, website, Point-of-sale, CRM, etc.) and the data science algorithms will be distributed across these systems. Hence it is important to understand the DS&A solution in the context of the enabling technology ecosystem and understand the pre-requisites to designing and delivering the solution.
And last but not the least, analytics may not be where your problem is. You may have the best customer acquisition models, but an un-informed sales force at the call center may result in fewer prospects turning customers. A call center head of a major bank once told me that the marketing department never kept them informed of campaigns. They found out as calls started pouring in!
Many a times, we hear that it is difficult to quantify the business benefits. Yes, the data scientist will be able to look at model accuracy, lift and predict the benefit. But to translate the predicted benefit into real business benefits on the ground, the DS&A expert has to factor in and think through the end to end process, associated technology systems and be focused on solving the business problem.