How grocers can improve demand forecast accuracy with generative AI

12.14.2023 | Mike Troy
Innovative copilots create productivity, precision, and predictability advantages

Retail supply chains will receive a major upgrade in 2024 as predictive and generative AI unlock wide-ranging benefits, especially in demand forecasting.

The catalysts for the imminent upgrade are powerful AI-based software copilots that give retailers the ability to tap into the rapidly advancing capabilities of predictive and generative AI in ways that weren’t previously possible. Demand forecasting is one area where the impact of copilots will be significant because of intensifying industry challenges following years of supply chain disruption caused by the pandemic, product shortages, and inflation.

Demand forecasting must deal with rapidly shifting consumer trends, the intricacy of retailers’ merchandising strategies, and varied store sizes with localized assortments. Adding to the complexity is the impact of e-commerce fulfillment and relentless financial pressures to minimize inventory investments while maximizing on-shelf availability.

The three Ps of Generative AI Copilots

The introduction of generative AI copilots to demand forecasting is a huge advancement because it impacts the three Ps of productivity, precision, and predictability. Looking specifically at the Demand Planner Copilot announced by SymphonyAI in October 2023, it simplifies access to predictive and generative AI to increase demand planners’ productivity, make their decision-making more precise, and improve the predictability of the outcomes of those decisions. This powerful combination is a welcome development for retailers and improves the lives of demand planners who gain the ability to:

  • Do more in less time: A boost in productivity is a core value proposition of the Demand Planner Copilot. One way is by removing the time-consuming task of gathering data and insights for day-to-day tasks like linking new items to related product demand, analyzing demand changes, and forecasting inaccuracies. Because the copilot works harder, demand planners can work smarter.
  • Improve accuracy and discover new opportunities: Demand forecasting can be a laborious process that can lead to discovering surface-level insights and the increased potential for inaccuracy. The Demand Planner Copilot fixes this issue because planners can analyze data in ways that previously may not have been attempted due to time constraints, providing a triple benefit of increased accuracy in less time and the potential to discover new areas of opportunity.
  • Increase confidence in decision-making and reliability of results: Demand planners using advanced tools to improve forecast accuracy increases retailers’ conviction that they will achieve planned performance. That’s because the Demand Planner Copilot provides transparency into recommendations and allows for manual intervention if or when needed. This eliminates the so-called black box effect that arises when users don’t have visibility into how AI arrived at its results.

Demand forecasting will never be an exact science due to the large number of variables that affect consumer demand combined with the increased complexity of supply chains. However, there is plenty of headroom to improve accuracy and speed through automation enabled by copilots.

Copilot 101

If 2023 was the year of generative AI innovation and discovery, 2024 will be the year of the copilot value creation as major technology companies and industry innovators have moved quickly to develop copilots. The beauty of a copilot is it lets humans engage with AI using natural language and offers easy-to-use capabilities that don’t require the advanced technical skills of a data scientist. As a result, a demand planner or forecast analyst using a thoughtfully designed, intuitive interface can query and analyze underlying data more efficiently and accurately.

This is why copilots are a breakthrough in the rapidly evolving world of predictive and generative AI because they solve real-world problems. Here are three quick examples:

  1. Improved performance with new items. New items drive growth, but accurately forecasting demand is notoriously difficult because, without any sales history, expectations for new item performance are based on past performance of items with similar attributes. This linking process is time-consuming and prone to inaccuracy and lost sales. A copilot can make a huge difference because a demand planner can use natural language to identify the optimal items for linking, and the copilot provides justification for the linkage. The result is a more accurate forecast that takes minutes rather than hours to develop.
  2. Identify unexpected volume shifts and underlying causes more quickly. A demand forecast may be created with ideal market conditions in mind, but real-world circumstances often cause actual results to differ materially from planned performance. Demand planners need to adjust fast to identify these differences and understand the source of disruption to prioritize and execute corrective actions. A copilot greatly accelerates this data-dependent process and offers insights into the root causes of forecast inaccuracies.
  3. Easier access to more data sources. A copilot simplifies access to a broader universe of data, which improves accuracy and creates a cascade of benefits related to optimal inventory levels, improved in-stock, and better financial results. The current reality for many demand planners is that they rely on disparate data sources in various formats that require considerable effort to unify before it is possible to query the data. The copilot increases the utility of data from multiple sources so demand planners can quickly extract the most important insights from the most relevant data.
Copilots create an uneven playing field

A rising tide may lift all boats, but there will be an uneven dispersion of benefits when the full force of the generative AI-powered innovation wave breaks over the retail industry in 2024. That’s because an innovation wave requires retailers to choose whether to ride it and how aggressively to do so. By contrast, it is a more passive exercise to benefit from a rising tide of increased consumer spending.

This is the choice in front of retailers in 2024 regarding copilots. To continue with the surfing metaphor, some retailers have seen the swell gathering momentum on the horizon and are ready to paddle in for the ride of their lives. Copilots will help them improve productivity and financial performance as the generative AI innovation wave rolls ashore.

For more helpful reading about generative AI and copilots, check out the following:

How generative AI accelerates retail innovation and increases productivity: A review of top use cases, new drivers of value creation and overlooked risks

Improving retail supply chains with generative AI

AI is making category management fun again

Latest Insights

06.12.2024 Webinar

Autonomous merchandising and the quest for shelf planning excellence

Retail / CPG Square Icon Svg
convenience fuel retailing Fuel Retail Analytics
05.17.2024 Blog

Shopper Analytics Give Convenience Retailers Fuel to Increase Loyalty and In-Store Sales

Retail / CPG Square Icon Svg
Kraft Heinz Strengthens Retailer Collaboration and Shopper Focus
Case study

Kraft Heinz Strengthens Retailer Collaboration and Shopper Focus

Retail / CPG Square Icon Svg