COVID-19 has turned the retail and e-commerce ecosystem upside down. While some businesses have seen a drastic decrease in sales, others have experienced a boom and marched to the forefront of the very competitive industry.
AI and machine learning technologies can play a key role when adapting to this new and less predictable reality. Organizations are leveraging their data to improve supply chain management, forecasting and to better understand buyer behavior.
We talked to data science experts in the industry and asked them about areas for growth potential, challenges and possible constraints when putting new data science solutions into practice.
“COVID-19 has radically transformed the retail industry and data is key in helping navigate this new reality. Shopify has over 1 million merchants worldwide, so when the pandemic started we developed a data task force to track the impact. Our Data team has been instrumental in providing timely insights to inform how we re-tool our products to help our merchants adapt. We’ve also put these insights, as well as data products, into the hands of our merchants. Whether they’re signing into their analytics dashboard to look at buyer behavior, downloading our AI-powered virtual assistant kit to automate their marketing, or receiving a funding offer from our ML-driven product Shopify Capital, data has empowered our merchants to make informed decisions during this critical time.” –Ella Hilal, Director of Data at Shopify
“Having accurate forecasting is crucial for retailers in order to avoid over and understocking. In fact, most retailers have invested a lot of time and effort in building accurate forecasting models in the past. Most of these models have been drastically affected given the new trends in data. This is where new machine learning approaches and data science skills can play a key role. Retailers must not only be creative on the adaptation or creation of new models, but also on the new data sources available to better forecast the uncertain short-term future.” –Maia Brenner, Data Scientist at Tryolabs
“Companies across multiple different sectors, especially retail, have had to adapt their business models due to COVID-19 if they are to remain successful. Supply chains have been disrupted directly, or due to increased pressures on carriers. During these times, it's more critical than ever to maintain consistency in the customer experience and product delivery amidst circumstances that are very different from what has been the case historically. Businesses that were running as a well-oiled machine and carefully optimized to generate revenue in a competitive retail market are totally thrown off due to the unpredictability and changing times.
Data science can help correct any business process that might rely on data and past experiences by correcting for and adopting signals to things that are more likely to be the case with COVID-19. In addition, there is an unanimous shift towards digital and async processes wherever feasible. So whether it's determining what products are feasible, or informing the leadership, or user about the current status, data science is essential. COVID-19 has also sped up the digital transformation by years, and led about sustained changes in user behavior. Data science can play an active and important role in helping the organization understand the implications of these sustained changes in customer behaviors, and the impact of changes in key assumptions that the business has made as part of its operation. Machine learnt solutions are much better suited to rapidly adapting to the new normal than hard-coded logic, and so companies that invest in data science find themselves much better positioned to leverage the changing customer behavior.” –Nishan Subedi, Head of Algorithms | VP of Technology at Overstock.com
“We are asking more from the data we have than ever before. Traditionally, data has played the role of providing insights and reports to executives in making decisions, whereas now we're starting to think of data as core to product offerings. This requires our data to be more available, accurate and usable. To realize this end state requires undertaking massive transformations of our technical stack towards responsibilities they haven't had in the past, and continue to be a big challenge in using data science successfully. In addition to this, there is still a gap in business and technical expertise employed to solve business challenges, and lack of communication between the two. Business owners don't communicate problems in ways that are broad and open enough for data scientists to innovate with solutions on, and data scientists aren't always great in explaining reasoning for technical decisions which prevents strong partnerships across different functions in the organizations. There are also inherent biases in the algorithms and constraints in systems. These limitations usually manifest as friction towards rapid experimentation and newer learnings that bias us towards previous 'tried and true ways' in reality are under-optimized due to changes in the competitive landscape. This friction poses the need for a big cultural transformation to better adopt data science, and thus a big challenge that companies face.” –Nishan Subedi, Head of Algorithms | VP of Technology at Overstock.com
“A major challenge in using data science is having all of your data in one place. One of the biggest changes we’ve implemented recently is the rebuilding of our Enterprise Data Warehouses (EDW) and building conformed data models for analysis. So how do we identify the right data for the questions being asked and being able to access that information? Another exciting change that we’re making is implementing more tools to allow the organization to self-serve verified and conformed data. Finally, communication of results is often the biggest challenge we face. As analysts, we often understand the data so well that we assume our audience just “gets it” too. We’ve been working towards better storytelling and Richard Branson’s saying that ‘if something can’t be explained on the back of an envelope, it’s rubbish’.” –Samantha Cvetkovski, Data Science Manager at Mindbody
“I believe the greatest challenge right now is regarding the investments in sectors that have been hard hit, such as brick and mortar. Those who have not yet made the digital transformation to adopt Machine Learning and still do not clearly understand the impact it could have on their organizations, are way more cautious about investing in these new technologies. This can increase the gap between the early adopters and the late ones. For instance, from a data standpoint those that didn't accumulate a lot of it in the past, will not be able to use it to do any kind of predictions.” –Maia Brenner, Data Scientist at Tryolabs
If you’d like to learn more about data science applications in the retail and e-commerce industry, join us for the Data Science Virtual Salon for Retail & E-Commerce from November 17-18. Free tickets are available for sponsored sessions or use promo code Roundtable20 to get a 20% discount on your General Admission Ticket.