What is Demand Forecasting?

Archived. This article has not been updated since the publish date above. The dynamic nature of information means that previously accurate content can become outdated or even obsolete over time. Readers are advised to exercise due diligence and cross-check any information found in this blog post before making decisions or adopting any practices based on said information.

What is demand forecasting?

• Demand forecasting is the process of projecting consumer demand (equating to future revenue). Specifically, it is projecting the assortment of products shoppers will buy using quantitative and qualitative data.
• Retailers are facing a trillion-dollar problem due to unavailable products at the time consumers demand them. Poor demand forecasting is causing companies to put the wrong product on the shelf, or an even bigger issue, to have in-store stock outages

How can Lakehouse for Retail enable demand forecasting?

• Lakehouse for Retail enables real-time access to data. Moving from batch-oriented access, analysis, and compute will allow data to be “always-on,” therefore driving real-time decisions and business intelligence. The Lakehouse uses technologies that include Delta, DLT, Autoloader, and Photon to enable organizations to make data available for real-time decisions.
• Lakehouse for Retail supports the largest data jobs at nearly real-time intervals. For example, customers are bringing nearly 400 million events per day from transactional log systems at 15-second intervals. Because of the disruption to reporting and analysis that occurs during data processing, most retail customers load data to their data warehouse during a nightly batch. Some companies are even loading data weekly or monthly.
• Lakehouse event-driven architecture provides a simpler method of ingesting and processing batch and streaming data than legacy approaches, such as lambda architectures. This architecture handles the change data capture and provides ACID compliance to transactions.
• DLT simplifies the creation of data pipelines and automatically builds in lineage to assist with ongoing management.
• The Lakehouse allows for true real-time stream ingestion of data, and even analytics on streaming data. Data warehouses require the extraction, transformation, loading, and then additional extraction from the data warehouse to perform any analytics.
• Photon provides record-setting query performance, enabling users to query even the largest data sets to power real-time decisions in BI tools.

REPORT

The agentic AI playbook for the enterprise

Read now

What data sources are needed for successful demand forecasting?

• Demand forecasting is an integral part of delivering products to consumers. To successfully deliver products to consumers, retailers must work with manufacturers in these operational processes and use cases: Sales and Operations Planning (S&OP), supply chain optimization, inventory control and optimization, and production scheduling. Data originates from several sources that feeds the demand forecasting models, including:
  • Point of sale: sales transaction is in the form of structured real-time data
  • Historical demand: Historical demand can be housed in enterprise resource planning systems (SAP) and usually structured in batch format
  • External data: Social feeds, news, and competitor information is usually in a real-time unstructured format
  • Click-stream: semi-structured in real-time
  • External data: even weather can be a factor in determining demand (think of tornado or hurricane)

Additional Resources

• Collaborating Across the Retail Value Chain with Data and AI
• Improving On-Shelf Availability for Items With AI Out of Stock Modeling
• How Starbucks forecasts demand at scale