You will be redirected back to your article in seconds
Skip to main content

Q&A: Impact Analytics Exec on Building More Accurate Retail Forecasting

As retail emerges from an unprecedented pandemic year, planning is a crucial component for driving recovery.

According to Melanie Casinelli, chief digital transformation officer at Impact Analytics, retailers need to move away from planning assortments and pricing based on the past and instead use data and artificial intelligence (AI) to more precisely predict future consumer demand and behavior.

Melanie Casinelli Impact Analytics
Melanie Casinelli, chief digital transformation officer at Impact Analytics Courtesy

Casinelli has been in retailers’ shoes. Before joining Impact Analytics, she held leadership roles in merchandising at companies including Urban Outfitters, Belk and Pandora Jewelry.

Sourcing Journal spoke to Casinelli about the difference with AI-powered forecasting and how retailers can plan for future disruption.

SJ: This past year has thrown multiple curveballs at the apparel retail industry. What role can artificially intelligent forecasting play in creating more agile, smarter supply chains to weather any further disruption?

MC: The retail environment is changing rapidly. Many people talk about when retail goes back to normal, but the reality is that retail will be discovering a new normal. Covid has exacerbated consumers’ expectations on being able to buy product when they want to buy, at the exact time they want to buy it, from the precise location they want to buy from. Thus, retailers need to evolve their level of consumer centricity through the investment of technology. The retailers that can remain agile and anticipatory are those that will win the consumer favor by being able to service them.

AI plus advanced analytics allow retailers to assimilate a large quantity of data from various sources and generate insights; AI connects relevant internal and external data for highly accurate demand forecasting. With the increased level of data and external factors, AI can forecast with high precision, anticipate disruptions and pick up on anomalies quickly. Machine learning (ML) leverages real-time data such as social media and the weather. There, a company leveraging machine learning will be able to react to a forecasted snowstorm by leveraging external data sources and the effects on its supply chain in real-time as opposed to a retailer using a traditional method where the system would not react until post the event occurring. This is just one of many examples to show the power that AI and ML have in changing an organization from reactionary to anticipatory, which in turn means increased customer satisfaction.

Related Stories

SJ: Excessive promotions have plagued apparel retail for a long time, but the pandemic worsened the problem. How does Impact Analytics’ software help retailers maximize their full-price sales?

MC: We help retailers optimize the lifetime margin of their product. Through elasticity modeling and AI forecasting, we can tell the retailers the price needed, at a specific time, in order to hit a specific sell-through target. What we have found is that many retailers rely heavily on what they did last year. This causes them to go too deep on some products, too heavy on others, along with suboptimal timing. By getting the timing and discount needed correct, the retailer can see substantial increases in lifetime margin and sell-through.

SJ: While some retailers faced pressure to clear out inventory due to decreased demand during Covid, others saw unexpected spikes in consumption. How can companies better prepare to avoid stockouts while also guarding against the risk of overstocks?

MC: We have not experienced a global pandemic like Covid-19 in over 100 years. Hence, it makes sense that this pandemic would stress the supply chains across many organizations, given that there is a large difference between organizations leveraging advanced analytics and ML algorithms in their demand forecasting compared to those that do not. Advanced Analytics and ML demand forecasting technologies can predict shifts in demand patterns (anomalies) quickly, whereas traditional forecasting methods struggle to keep up. My suggestion to retailers would be to invest in demand forecasting technology that leverages machine learning and advanced analytics. The rapid pattern recognition and ability to account for external demand influencing factors will keep you ahead of your competitors still leveraging traditional methods.

SJ: What differentiates Impact Analytics from other AI planning solutions for retailers?

MC: Two of the biggest things that differentiate Impact from other AI planning solutions are forecasting accuracy and agility. We are continuously tweaking our models to increase the level of accuracy in our forecasting. For example, when Covid began, we knew we were up against an unprecedented event that would have a significant impact on our retailers. We immediately got our arms around external factors that we could pull into our model to strengthen the output. These external factors have given us the ability to get our forecasting error rate to below 10 percent during Covid. With this level of accuracy, we can give our clients the ability to automate a large part of their workflow because they can trust the forecast accuracy. Many of our competitors treat their clients as a one-size-fits-all model. We know that each retailer is different and will have unique configuration needs. Therefore, we have built agile products that can be configured to optimize all retailers’ businesses.