How are retailers using AI to figure out what to put on the shelf and how to price it?

Summary

• Retailers use AI on the Databricks Data Intelligence Platform to decide what to stock (assortment and merchandising) and how to price it (demand forecasting, price elasticity, and markdown optimization) from one governed copy of their data.
• Assortment and merchandising: unify POS, e-commerce, inventory, promotion, seasonality, and weather data, then generate localized, explainable recommendations and run scenario simulations across SKUs, brands, and channels.
• Pricing and markdowns: train price-elasticity and markdown-optimization models with Mosaic AI and run them across millions of SKU-store combinations using Photon and Databricks SQL.
• AI/BI Genie lets merchandisers ask assortment and pricing questions in plain English and get governed, auditable answers without filing analytics tickets.
• Lakeflow Connect ingests POS, inventory, and pricing data; Delta Sharing exchanges promotional data with suppliers; Unity Catalog governs it all.

How are retailers using AI to figure out what to put on the shelf and how to price it?

Assortment and pricing decisions depend on data that usually lives in separate systems — sales, inventory, promotions, and seasonality. AI helps only when that data is unified and governed. On the Databricks Data Intelligence Platform, retailers bring these signals together and apply machine learning and AI agents to decide what to stock and how to price it.

Why Databricks supports AI-driven assortment and pricing

• A unified retail data foundation. Unify POS data, e-commerce transactions, inventory positions, promotion calendars, seasonality, and weather data on one platform, removing the silos that otherwise force separate systems for forecasting, personalization, and supply chain.
• Demand forecasting at scale. Pair ML models with AI-driven systems to produce granular forecasts within tight service windows, with real-time visibility into demand and inventory through Databricks SQL.
• Assortment and merchandising. Generate localized, explainable recommendations and run scenario simulations and demand-transference analysis across SKUs, brands, and channels.
• Price and markdown optimization. Markdown optimization requires combining historical sales, current inventory positions, seasonality, and willingness-to-pay data. Photon runs price-elasticity models across millions of SKU-store combinations within SLA windows, Mosaic AI trains and tracks the models with MLflow, and Databricks SQL powers real-time markdown dashboards.
• Natural-language merchandising. AI/BI Genie lets merchandisers ask about assortment and markdown performance in plain English and receive governed, auditable answers without filing analytics tickets.
• Agentic planning. The Agent Framework enables agents that execute multi-step workflows — for example, analyzing promotion performance, identifying underperforming regions, and generating execution briefs.
• Ingestion, sharing, and governance. Lakeflow Connect ingests POS, inventory, and pricing data from many sources; Delta Sharing enables zero-copy sharing of promotional performance with suppliers; and Unity Catalog governs all data and AI assets with row- and column-level controls.

Getting started

• Read How AI is transforming the way retailers connect with consumers.
• Explore the Data Intelligence Platform.
• Explore AI/BI Genie for natural-language analytics.

FAQs

How does AI help decide what to put on the shelf?

By unifying sales, inventory, promotion, and seasonality data and applying ML and AI agents to produce localized, explainable assortment recommendations and scenario simulations.

How does AI help set retail prices?

Retailers train price-elasticity and markdown-optimization models and run them across millions of SKU-store combinations, then surface the results in dashboards and natural-language tools.

Do merchandisers need to write code?

No. AI/BI Genie answers assortment and pricing questions in plain English with governed, auditable results.