Successful retailers are data-driven retailers. The task of predicting consumer behavior, with an eye toward understanding individual customers, involves working with dozens of variables and data that is often incomplete, ambiguous, and organized for transaction processing and not analytics.
The promise of data science, artificial intelligence (AI) and machine learning is for these techniques to help spot trends and patterns with sufficient speed and accuracy for retailers to be able to adjust floor-sets, advertising, discounting, and product mix as quickly as possible. But just as data scientists are becoming more commonplace, antagonism is growing between expectations about how quickly algorithms and advanced mathematical techniques can transform retailers, and the reality that data science is, well, science.
Just as with the physical sciences, there are many dead ends and failures in data science. It takes a visionary retail senior manager to balance expectations with the patience to allow these methods to iterate and mature. One thing that can reduce the risk, complexity, and timelines associated with applying data science to the retail business is to recognize that, regardless of the mathematical, AI, or machine learning techniques used, 80% of the effort involves connecting the data.
Recently, at the Women in Analytics Conference in Columbus, OH, which drew over 450 data analytics and data science professionals, many of the presentations featured breakthrough insights and compelling visualizations. But there was a common theme: behind the relative glory of the cutting-edge science were stories about never-ending data integration, cleaning, transformation, and loading. The reality is that if retailers could solve the data integration problem, and reduce both the effort and latency involved, they’d have more resources and time to explore innovative data-driven techniques for enhanced decision-making, from supply chain, to market basket analysis, to campaign management.
Retail organizations seek to understand their customers’ relationship with products, stores, advertisements, and manufacturers, and the transactions that tie them all together. To do this, they must be able to express information from different, siloed applications and systems as entities and relationships. If data scientists are enabled to work directly with entities and relationships, a major chunk of the data preparation labor can be reduced.
Furthermore, there’s a compelling business case to persist these entities separate from the underlying data sources from which they are drawn. As soon as retailers recognize that data access, integration, and preparation comprise most of the work associated with data science, AI, and machine learning, the topic of data governance comes up. Data governance helps ensure that high-integrity data is available, usable, and secure throughout the organization and associated business processes. At the heart of data governance is a mutual understanding of the data model by a retailer’s key leaders and stakeholders. However, for today’s retailer, the data model can’t be some brittle monolith. Retailers are facing business challenges that evolve faster than their IT departments can respond. To meet these increasingly tight deadlines and expectations, the way they approach data governance has to shift.
We need to use technology to connect our data faster so that we can start using data science, AI, and machine learning to get to the insights we want – the good stuff. Retailers expect data analytics to provide better insights into customer interests and buying habits. Data scientists should not be wasting their time trying to pull data together from a bunch of disparate sources, and instead of wrangling raw data they should be using their talents to work with the entities and relationships to discover superior insights. THAT is what retailers need and THAT is the promise of data analytics.
From investment banking to craft distilling to technology, Megan Kvamme has provided leadership to a variety of industries throughout her career. She is the founder of several women-owned businesses, including the technology start-up FactGem, where she is also CEO.