Poor grocery demand forecasting causes more waste than you may think.
According to one statistic, grocery retailers in the United States discard 10% of the country’s yearly food production of around 44 billion pounds. It’s not just terrible for the environment (food waste is a significant source of carbon emissions), but it’s also expensive for retailers. According to Retail Insights, food and grocery businesses lose up to 8% of their income due to insufficient inventory availability.
Entrepreneurs Euro Wang and Jack Solomon claim they saw the micro-level consequences of the forecasting issue directly at their local store, which often ran out of their favorite guacamole.
“It turns out that even the largest retailers struggle to predict future demand and frequently overstock and understock inventory,” Wang said to Eltrys via email. “With more extreme weather in recent years, there has been an increase in fresh food supply shortages.” This emphasizes the need for proper distribution of the limited supply. Furthermore, inflationary pressures and rises in labor expenses have been jeopardizing grocers’ profits.”
Inspired to utilize technology to solve the issue, Wang and Solomon co-founded Guac, a platform that uses artificial intelligence to anticipate how many products grocers will sell per day at a specific shop location. Guac recently secured $2.3 million in a seed round that also included Y Combinator and Collaborative Fund.
“Food waste and food security are issues that Jack and I care deeply about, and we were really excited about an opportunity to actually address food waste at its core,” Wang said in a statement.
Wang previously worked at Boston Consulting Group, where Solomon studied artificial intelligence for food logistics. Both received undergraduate degrees from Oxford University, where they met.
Wang, Solomon, and Guac’s two engineers create proprietary algorithms that predict order amounts for supermarket products, taking into consideration factors such as weather, sports events and betting chances, and even Spotify listening data, in an attempt to capture customer spending behavior. Guac clients have shelf life, minimum order quantities, promotions, and supplier lead times recommended to them linked into their current inventory ordering software and processes.
“Traditionally, forecasting is done using Excel formulas or simple regression models,” Wang told me. “However, for fresh food that spoils rapidly, you need something better… We can pinpoint which real-world factors influence variations in demand since we employ so many external variables.”
Guac is far from the only company attempting to anticipate food demand. Crisp, for example, offers an open data platform for each link in the food supply chain, while Freshflow is developing an AI-powered forecasting tool to assist merchants in optimizing stock replenishment of fresh, perishable commodities.
However, Wang asserts that Guac stands out for its dedication to openness and meticulous fine-tuning of forecasting algorithms.
“Our machine learning model isn’t like a black box that mysteriously predicts a 20% increase in demand; instead, we tell our customers things like, ‘This 20% increase is because there’s a conference happening nearby,'” Wang told me. “Even if a retailer is already using machine learning, we can improve their forecasting because we have access to a much larger set of external datasets.” We really observe the prediction inaccuracy increase when we remove our unique external factors and simply use the core datasets (e.g., weather and public holidays).”
Some early adopters seem to believe that Guac can provide value. The startup is collaborating with merchants in North America, Europe, and the Middle East, including an undisclosed supermarket chain with 300 or so stores. Guac is also already making money and plans to increase its technical staff in the next year.
“The grocery industry is fairly resistant to economic downturns,” Wang told Reuters. “Everyone needs to eat, and when the economy slows, people buy more groceries because they eat out less.” Furthermore, the epidemic accelerated grocery shop digitalization, allowing us to more seamlessly link our forecasts with consumers’ systems. Concerning the epidemic, buyers acted extremely differently during the pandemic, making it far more difficult for grocers to estimate future demand based just on the previous three years of historical sales data. With our approach, we can account for how the pandemic distorted sales data in 2020 and 2021—an even for the epidemic’s aftereffects.”
