84.51° Data science and retail analytics in action

At 84.51°, it’s not just our data that sets us apart – it’s putting what we know about the data into action. By applying our retail data science, insights, and media activation solutions, we help Kroger, brands and other retail businesses develop strategies that help them optimize operations so they can meet their customers where they are with what they need.

Retail analytics power a better shopper experience

Our first-party retail data comes from over 60 million U.S. households and is sourced through the Kroger Plus loyalty card program. Every time a customer swipes their loyalty card at Kroger it does two things: It gets them a discount on the items they buy, and it lets us know what they like or dislike from Kroger.

The more sales data that’s collected, the more we understand what customers really want. These valuable insights enable us to build meaningful relationships with customers. It allows us to tailor our direct communications to show our customers we see them and understand them. For example, when we analyze the data, we know that it's a waste of time marketing a honey-roasted ham to someone who’s a vegetarian. We also know that giving someone a $3 coupon, instead of a $2 one, can allow that family that’s trying to make ends meet add another item to their basket to make their meal more filling.

Sales data also helps retailers how to optimally run the business, such as how to tailor a store to meet the needs of a community or which products will be relevant to the community. In addition to helping identify and understand each community’s unique character and preferences, it’s also a valuable tool for optimizing the entire retail ecosystem – from streamlining the supply chain and improving inventory management to setting optimal prices.

Data science at work in retail analytics

Once we have all of the shopping trips for all of our households aggregated, we then apply data science and retail analytics to improve and optimize the shopping experience. Data science influences the entire retail ecosystem. It enables us to (1) look back to extract insights from data sources such as customer feedback, financial performance, and operational metrics, and diagnose any unexpected behavior (2) develop predictive models such as forecasting and recommender systems (2) prescribe and optimize business decisions based on predictive models and business objectives. It’s a continuously evolving discipline, especially with the advent of artificial intelligence (AI). Researchers have been talking about and experimenting with AI since the 1950s, and the retail industry has been using AI for more than a decade.

Machine learning is a subset of data science that focuses on building models and algorithms that learn patterns and make predictions. It helps people analyze massive amounts of data from multiple sources and find connections between data faster than humans can do. Some of its tools and functions include:

• Supervised learning: learns to recognize patterns and relationships between inputs and outputs, allowing it to make predictions on new, unseen data, e.g. forecasting models

• Reinforcement learning: machines to learn and make decisions by interacting with their environment, receiving feedback, and optimizing their actions to achieve specific goals, e.g. real time analysis

• Optimization: computational procedures used to find the best solution or outcome among a set of possible choices by maximizing or minimizing a specific objective, e.g price optimization

• Generative AI: create new content, such as images, text, or music, by learning patterns from existing data, e.g. new product recommendation

Technology advances fueling retail analytics

Our data scientists’ goal is to leverage advancements in technology to build trusted and responsible algorithms to create more valuable experiences for our shoppers. Such algorithms typically involves utilizing multiple of the tools listed above. We incorporate such advanced algorithms into multiple areas, such as:

• Augmented or autonomous processing where we use predictive models (such as forecasting models) and prescriptive sciences (such as optimization) to augment judgement in operations and merchandising. This includes use cases like store ordering, and warehouse optimization.

• Intelligent media product offerings where we deploy in analytical product offers that are complete with model monitoring and interpretability. We see this come to life through our Kroger Precision Marketing (KPM) platform, where we have optimized audiences that brands can utilize to maximize ROI for their campaigns.

• Robust insights or decision intelligence where our advanced analytics and machine learning is embedded throughout the enterprise to support decisions – such as our customized assortment and shelf optimization sciences.

• All around personalization, using AI to create personalized, relevant, unique experiences for our shoppers whether it is the prices they pay, their experience on the Kroger app, the content and creative offers they receive, or the products they see.

• Manufacturing where we are utilizing latest AI agentic framework to enable us to optimize formulations and rapidly speedup development of new and innovative products to satisfy evolving customer taste and preferences

We already have AI and machine learning embedded in many places across the Kroger ecosystem and has demonstrated tangible business value. While the buzz around AI is loud, generating value out of such algorithms is hard – something we have successfully demonstrated time and again, and in multiple areas of the business. Across merchandising, for instance, there are optimization and forecasting sciences to drive pricing, promotion and assortment decisions. These advanced machine learning algorithms are fed into systems and processes across Kroger to create a better customer experience and make the lives of merchants easier. Across the digital e-commerce experience, deep learning algorithms make it easier for customers to save time and money while ensuring they receive the most relevant products and offers for them. And we are finding ways to elevate and amplify the voice of the customer through improved sentiment analysis.

We are also leveraging AI across our stores, supply chain (transportation and wearhouses), HR, and customer experience. We will continue to invest in new capabilities and create scalable and democratized frameworks for AI, to embed AI in all we do.

For example, Sage is a personalized AI virtual assistant that puts labor data and metrics at the fingertips of store leaders. VROOM is a machine learning model that optimizes outbound truck routing by trying out thousands of possible route options and building the optimal route. IRIS leverages data intelligence to improve sell through rate based on a specified sell through window, such as seasonal or holiday. And Quevision leverages AI to predict traffic in checkout lanes, allowing for better planning of staffing to reduce customer wait times, creating a better associate and customer experience. We also continue to build out the next generation of optimization sciences to optimize price & promotion together to improve category performance and create a division-level executable sales plan.

Optimizing supply chains with retail analytics

Another example of the power of data science and retail analytics lies in supply chain transportation optimization. In the retail industry, current supply chain movements are siloed, disconnected from the business, and handled manually. Also, they Applying retail analytics tools can help organizations optimize all supply chain movements for a future where every transportation decision is connected to the customer , overall business and its environmental impact.

Route optimization is a retail industry challenge, and as size increases, it becomes very difficult to solve this problem fast. Vehicle routing faces a myriad of constraints including time windows of locations, resources available (vehicles and labor), vehicle capacity, types of goods (frozen or perishable items have a shelf life), and the traffic and roads that are many time uncertain and hard to predict in advance.

The typical routing system in place today is a manual process with lots of potential for evolution. But the data science-based solution is fully automated and uses cutting-edge science to optimize the entire fleet. Because retailers spend significant amounts of money on outbound transportation annually, every percentage of improvement in truck routing results in massive cost savings and reducing environmental impact, such as reducing gas consumption. This solution also offers ethical benefits as it contributes to a more sustainable supply chain.

“The end seems bit abrupt. Should we say something like: we have demonstrated that such solution when scaled across the retail business has the potential to significantly enhance transportation and adds significant value. For comparison, for a US transportation industry, which is more than $40B in size and contributes to over 300 billion miles every year mere 5% saving through optimization present a value proposition of $2B in saving and 15B less miles”

Learn more about retail analytics in action

To learn more about how data science and retail analytics deliver performance for retailers and brands – and value for shoppers – visit our Knowledge Hub.