Success in Retail Machine Learning
With Amazon ($228M), Alibaba ($300M), Google ($130M), and Microsoft ($75M) spending large sums on Machine Learning, how will retailers keep up?
Many companies have launched small data science teams in response and we are seeing some consistent issues with these teams.
Data scientists typically don’t understand retail and where/how to best leverage machine learning to enable a broad retail transformation.
Business leaders don’t understand data science well enough to understand how to leverage it as a tool in transforming their organizations. In addition, business organizations tend to be very comfortable with their “old math” and their use of Excel as their primary data science tool.
As a result we are seeing a larger disconnect between data science and business than we saw 25 years ago between information technology and business.
So how do retailers go about solving these issues?
First, by holding retail executives accountable for introducing and testing machine learning into their areas of responsibility.
“Any business task or consumer experience that can be automated, should be automated.” – Amazon
“Much of what we do with machine learning happens beneath the surface…. Quietly but meaningfully improving core operations.” – Jeff Bezos, 2017 letter to shareholders.
Machine learning can be successfully applied to product development (trend), merchandising optimization & automation (inventory, price, assortment, promotion), supply chain automation & optimization, fulfillment automation & optimization, returns management automation, store operations automation & optimization, contact center automation & optimization, finance automation, and HR automation.
Second, retailers must constantly survey the global machine learning solutions landscape for innovative machine learning solutions (not just retail machine learning solutions).
While IBM, Oracle, SAP, Salesforce and Microsoft all market that their latest retail solutions include machine learning, it is important to also look beyond them. These solution providers have a large revenue base to protect so tend to be slower, more incremental, and more expensive than other solutions).
Also look at machine learning startups with solutions that can transform various business areas and processes. A great example is www.crowdanalytix.com.
Machine learning solution providers also need make their solutions less “black box” (that only data scientists understand) to a set of tools that are easier for retail business leaders to understand and use. As this happens machine learning adoption will grow and new retail machine learning use cases will be imagined and implemented.
And, it is time for retailers to establish a machine learning strategy & plans just as 25 years ago they established information technology strategy & plans.