Amazon Supply Chain and Logistics

Maximize business value with the AWS Supply Chain demand planning process

In today’s complex global supply chain landscape, accurate forecasting is crucial—but it’s not enough on its own. Organizations have invested in building advanced analytics and Machine Learning (ML) capabilities to improve forecast accuracy and drive optimal inventory. Despite these extensive efforts, the inventory-to-sales ratio has been increasing since 2021, reflecting that organizations have been overstocking to account for demand and supply fluctuations. This fact suggests a missing link between improving forecasts and achieving tangible business value.

During periods of high supply chain disruption, collaboration and business expertise become even more critical. Without effective stakeholder communication, even the most sophisticated forecasts can quickly become obsolete, diminishing business value and leaving organizations vulnerable to rapid market changes.

The key to unlocking true supply chain value lies in combining cutting-edge forecasting with effective stakeholder collaboration. This post explores how AWS Supply Chain’s innovative solution empowers organizations to not only leverage ML for good quality forecasts but also to drive meaningful collaboration, capturing vital business insights to create demand plans that truly maximize economic value in any market condition.

Building high value demand plans

Traditional demand planning is a single-step process, focusing solely on generating the most accurate forecast possible using historical data. In highly dynamic environments like today, such a process yields limited success, as past events may not have same effect in future and additional expert inputs are also important.

AWS Supply Chain recognizes that truly valuable demand planning requires a two-step approach.

Step 1: Build strong baseline forecast using ML

In the first step, demand planners leverage AWS Supply Chain’s advanced analytics and ML capabilities to generate a baseline forecast. The forecast model analyzer runs on organization’s datasets and helps planners select the best-fit ML model, incorporating external factors and product transitions to maximize initial accuracy. For example, algorithms like DeepAR+ works with large datasets containing hundreds of feature time series and accepts forward-looking related time series and item metadata to generate forecast output.

Step 2: Drive stakeholder collaboration to build consensus-based demand plan

The second, crucial step involves collaborative refinement. Planners use AWS Supply Chain’s intuitive user experience to share the baseline forecast with stakeholders, gather insights, and capture critical business assumptions. This process allows for necessary overrides and adjustments, ensuring the final plan reflects a holistic, aligned view of the business.

Traditionally, assessment of demand planning processes adding business value relies on accuracy metrics like Mean Absolute Percentage Error (MAPE) or Weighted Absolute Percentage Error (WAPE). While these metrics remain essential for evaluating overall performance, they fall short in measuring the value added in each step of the two-step process. They focus solely on forecast deviation from actuals, overlooking the critical inputs and adjustments made during collaboration.

This is where the concept of Forecast Value Add (FVA) becomes essential. FVA measures the effectiveness of machine forecasts and human intervention in the forecasting process discretely, helping organizations identify which activities truly add value and which may introduce noise or bias. By analyzing FVA, companies can optimize their demand planning process, focusing efforts where they matter most. For example, Tempur Sealy International, the world’s bedding provider and RS components have successfully implemented FVA and benefited from understanding which inputs were truly improving their forecasts rather than creating noise.

AWS Supply Chain’s comprehensive solution facilitates this two-step process and provides reporting and analytical tools that can be used to compare actuals to naïve and consensus forecast and calculate FVA to assess value add at each step of the two-step process, enabling organizations to create demand plans that truly maximize business value beyond simple accuracy metrics.

Do not just forecast — forge your future

Demand planning is not just about accuracy— it is about creating real business value. AWS Supply Chain’s two-step approach aids this process, combining cutting-edge ML forecasting with powerful collaboration tools. By leveraging both data-driven insights and human expertise, organizations create demand plans that not only predict the future but shape it. In today’s volatile market, this is not just an advantage—it is a necessity. With AWS Supply Chain, turn your demand planning into a strategic advantage that propels your business forward. Begin your journey by:

  1. Evaluating your current process: Assess how your organization balances forecasting accuracy with collaborative input. Are you truly maximizing the value of both?
  2. Exploring AWS Supply Chain: Schedule a demo with your account management to see how our integrated ML forecasting and collaboration tools can transform your demand planning.
  3. Get a technical overview: Explore the AWS Workshop Studio for a self-paced technical walkthrough. You’ll learn how to create an instance, ingest data, navigate the user interface, create insights, and generate demand plans.
Amit Shah

Amit Shah

Amit Shah is a Principal Specialist Solutions Architect for AWS Supply Chain. In his role, he works with Supply Chain executives and Technical architects to help understand customer problems and transform customer supply chain to help achieve the intended business outcomes. He is Lean Six Sigma Black Belt certified and has over 18 years of industry experience in driving business and process transformation through the breadth of Fortune 500 industry spectrum starting from Med-Tech manufacturing, Technology, E-commerce and Cloud Infrastructure . Amit is based out of Greater San Francisco Bay Area.