Artificial Intelligence and Machine Learning in demand planning

This article summarizes some of the current use of Analytics tools based on Artificial Intelligence (AI) and Machine Learning (ML) in demand planning, highlighting the potential to improve accuracy in an increasingly complex environment, with diversified portfolios, channels of varied distribution and fierce competition.

It is already possible to observe the improvement in accuracy through the increase in the volume of data treated and variables considered in the models, allowing the identification of imperceptible patterns to the “human eye” and, also, automation of the treatment of data from the sales series, improving the baseline and making it possible to use more granular sales information.

In addition to improving the accuracy of the monthly tactical plan, the use of AI and ML in demand planning can also help identify long-term consumption trends, guiding portfolio definition and new product development. Additionally, it can improve short-term replenishment mechanisms with demand sensing techniques.

Another potential benefit is the ability to simulate future scenarios based on different sets of variables, such as changes in the economy, weather patterns or changes in consumer preferences. These simulations can help companies quickly adapt to market changes and make better decisions about resource allocation.

Below, I summarize the potential benefits and cite examples of companies that are already embracing AI and ML in their planning activities:

1. Increased accuracy of forecasts: identification of complex patterns and inclusion of a greater number of exogenous variables, improving the accuracy of demand forecasts. Amazon has developed algorithms to help it deal with varying seasonal patterns, estimate the impact of specific promotions and events, such as product launches, as well as anticipate the movement of products to its fulfillment centers;
2. Automation of data processing: automating data processing of sales series allows the use of more granular information, which can lead to better forecasts and adjustments in the baseline. Walmart automates the processing of sales data from its stores, allowing the planning team to focus on analyzing trends and making strategic decisions;
3. Identification of long-term consumption trends: advanced data analysis can help companies identify emerging trends, guiding portfolio definition and new product development. Pharmapacks has developed algorithms to identify the growing demand for natural and organic products, directing efforts to develop and promote products that meet this trend;
4. Improvement of short-term replenishment mechanisms: Demand sensing techniques can be improved with the help of AI and ML, optimizing inventory replenishment and reducing excesses or shortages of products. Zara uses demand sensing techniques to identify rapid changes in demand for different styles and sizes, optimizing inventory replenishment and reducing the amount of unsold items;
5. Simulation of future scenarios: simulating scenarios based on different sets of variables, such as changes in the economy or weather patterns, helps in making strategic decisions. Tesla simulates scenarios related to changes in government policies, such as tax breaks for electric vehicles, and adjusts its production and distribution strategy accordingly;
6. Optimization of marketing and promotion strategies: An in-depth understanding of the factors that drive sales can help companies optimize their marketing and promotion strategies, increasing the effectiveness of these actions. Coca-Cola and P&G use AI and ML to analyze the impact of different promotions and marketing campaigns on consumer behavior, allowing companies to direct their resources towards more effective actions;
7. Adapt quickly to changes in the market: With advanced analytics and more accurate forecasts, companies can adapt faster to changes in the market and allocate resources more efficiently. Mercks uses AI and ML to monitor and predict changing market conditions, such as the approval of competing new drugs, and quickly adapts its launch and pricing strategy to remain competitive.

These are just a few examples of how AI and ML have been successfully applied in demand planning and market understanding. However, the adoption of AI and ML in demand planning faces significant barriers, such as low maturity of the current process, lack of technical knowledge and integration in the chain to obtain data. 

Overcoming these challenges involves improving the maturity of the S&OP and IBP process, training the team in classic sales forecasting models and analytics tools, and advancing collaboration in the supply chain. Implementing a unified data platform that allows for the collection, processing and analysis of information from multiple sources is crucial to the success of these initiatives.

– McKinsey – How Amazon is using machine learning in its demand forecasting

– Walmart – Walmart establishes strategic partnership with Microsoft to further accelerate digital innovation in retail

– Analytics India Magazine – Procter Gamble leveraged analytics tide FMCG slowdown

– Forbes – How Coca-Cola uses AI to drive success

– Analytics India Magazine – Unilever saves 500 million euros using AI analytics

– Forbes – Zara uses supply chain to win again

– Merck Group – AI in supply chain

– Medium – How Tesla is using AI and big data analytics in its self-driving cars

– Analytics Steps – How Tesla is making use of artificial intelligence in its operations