Data signals for AI‑based decision making in retail and CPG

As excitement grows around the practical use of Generative AI, chatbots and Agentic AI in grocery and retail, it’s important to shine a light on the data innovations that power these models.

It’s remarkable that the humble grocery receipt has evolved to eventually become the fuel that drives some of retail’s toughest decisions at scale: who to sell to, what to sell, where to sell and at what price. Insight that generates customer, product and location signals underpins decades of innovation in segmentation, ranging, personalisation and pricing - and the journey continues as we push further.

The improvement cycle: enrich and combine

In the digital data capture explosion spanning browsing behaviour, app usage, images, traffic patterns and more AI has rapidly improved our ability to process and enrich these datasets − whether tagging images or extracting nuanced sentiment from social media

These advances can then be used to enhance our customer, product and location data, giving us deeper, more granular understanding.

Using pseudonymised data, we can identify:

• Shopping behaviour across different merchants
• Product features and ingredients
• Competitor activity around specific stores

When these enriched attributes are recombined, they form new signals, which are indicators of needs, preferences and experiences that unlock smarter decisions. This creates opportunities such as:

• Competitive Threat Evaluation
  By combining retail preference research with store‑level footfall data, we can spot high‑risk stores early and recommend actions to protect revenue and market share.
• Market‑aware Customer Decision Trees
  Using market transactions and substitutability science, we can build Customer Decision Trees that highlight true range gaps.
• New Product Launches
  Linking product attributes with behavioural data and social sentiment increases launch success for both national brands and private labels.

These examples show how diverse data, when connected, adds powerful context that improves decision making in retail’s complex landscape.

What AI in grocery retail promises

For customers, AI will make decisions faster (real‑time personalised offers), smarter (agents weighing price, brand and channel) and more accessible (chatbots that guide shopping choices).

For businesses, AI becomes a strategic asset − forecasting shifts, flagging risks, detecting competitive changes and anticipating supply chain shocks.

Achieving this requires the right data foundation. RAG and Agentic RAG frameworks depend on fast, accurate search, retrieval and consolidation of signals. If the underlying data is weak, millions of pounds can be wasted with little return.

At dunnhumby, we’re building this foundation through a dynamic, multi‑source data ecosystem that links multiple domains. This generates new attributes and fresh context relevant to today’s world—from customer needs and emotions to technology usage, regulation and emerging models like agentic commerce.

Our success is powered by:

• A strong network of reputable data partners
• Deep understanding of global retail and CPG challenges
• Decades of expertise in customer behaviour and data‑driven decisioning
• A world‑class AI, Data Science and Engineering team ensuring robust taxonomies, hierarchies and quality standards

Data is the fuel of AI engines, and retailers need more of it than ever and presented in the right way for the agentic era.