Advertisement
01/07/2021

How AI Can Solve Retailers’ Assortment Issues

How AI Can Solve Retailers’ Assortment Issues
Brian Ross, President of Precima, a Nielsen company.

Retailers of food and consumables begin 2021 with an operational challenge unlike any they have faced previously. Accurately forecasting shopper demand by category, brand and channel has been disrupted due to challenging comparisons with prior year results when the onset of the COVID-19 pandemic led to erratic and irrational spending behaviors. To better understand how retailers can address these unusual circumstances to make optimal assortment decisions, Progressive Grocer spoke with Brian Ross, the president of Precima, a global retail strategy and analytics company that was acquired by Nielsen in 2020.

Progressive Grocer: Given the disruption in demand we saw throughout 2020, what do you think are the main challenges and opportunities when it comes to assortment planning in 2021?

Brian Ross: Assortment management is a challenge in the best of times and requires the constant deployment of best practices. The COVID-19 pandemic has elevated assortment management to a new level as demand signals have been shattered by unpredictable shopper behavior across categories. Massive disruptions mean supply chains struggle to catch up with shortages caused by both consumer stockpiling and production challenges, impacting the entire industry in profound ways. Retailers and their suppliers responded heroically, but out-of-stocks and supply-chain complications have persisted in many categories. I fully expect this situation to dominate the thinking of merchandisers for most if not all of this year. Beyond 2021, there will be a significant SKU rationalization as retailers do a postmortem review of assortment planning during the pandemic, which will likely result in a cutting of many existing SKUs and the addition of many new ones.

PG: How can artificial intelligence help with assortment planning?

BR: The disruptions that we saw at the shelf in 2020 are not evidence that existing merchandising and assortment planning solutions have failed - rather it was a reflection of the extraordinary changes in consumer demand, preference and buying habits of consumers in the face of the pandemic that could not have been predicted based on previous behavior using standard methods. It was a true "black swan" event and to understand the impacts and how to respond requires powerful analytics to understand rapidly changing and evolving shopper behaviors.

This is where AI comes in. Applying AI to item-level transactional data, customer data and market data enables understanding of changing consumer buying behavior in face of the pandemic from purchase frequency and trip types to preferred days and hours of shopping through to the categories, items and brands purchased. Only through AI can these insights be enabled at the individual customer-level and at the moment, updated dynamically over time as we continue through the pandemic and beyond to keep pace with the ever-changing needs and demands of customers.

PG: Have any best practices emerged for AI food retail product assortment planning?

BR: A few things. Food retailers need to start by understanding the true consumer behavior change over the past year. This will reveal implications and opportunities for assortment. Given the “black-swan” nature of COVID-19, it is critical not to solely rely on traditional assortment modeling and algorithms. An adjusted approach will help understand item preferences and substitution and demand-transfer. Retailers need to assess assortments more frequently. In periods of instability they should review and adjust their plans as behaviors and circumstances change, not according to a fixed schedule. This also requires closer collaboration with vendors, especially on the demand and supply forecasts.

PG: What are the main things food retailers need to know about AI assortment planning going into 2021?

BR: Assortments will certainly look different in many categories. There will likely be fewer varieties, package sizes and brands. It is also probable that a few new brands that gained a toehold during the crisis will remain in the mix. This presents an opportunity for retailers that have struggled with SKU proliferation to smartly rationalize the assortment to key items in key categories that appeal to key customers.

This rationalization can only be optimized with the help of AI-based solutions that automate many of the easy decisions, leaving the most critical ones to the merchandisers. AI helps retailers truly understand the impacts of everything from how delisting a product that could result in profitable customers shopping elsewhere to what synergies there are between products that result in bigger market baskets. And it does it in near real time so there is little latency in perusing the opportunities identified.