From Our Print Edition

Brian Harris

Category management revolutionized grocery retailers’ purchasing operations in the late 1980s and 1990s. Today, advances in artificial intelligence (AI) are poised to similarly transform how grocers manage their categories, with profound implications for frozen and refrigerated foods and beverages.

Count Dr. Brian Harris — widely known as the “father of category management” — among the growing number of grocers and industry experts who recognize AI’s game-changing potential. Based in Los Angeles, Harris co-founded IntentAi, Inc. in 2024 to bring the next generation of category management to life. The firm claims its technology platform is the first to fully integrate AI into the category management process.

FROM THERE TO HERE

The consumer market conditions under which Harris developed category management in the 1980s are starkly different today. Then, shoppers were predictable, trading channels weren’t blurred and weekly syndicated data gave grocers all the information they needed.

“That world no longer exists,” observes Harris. “Today’s consumer shops across physical stores, e-commerce, clickand-collect, quick commerce and social platforms — often within the same week and sometimes within the same purchase journey. Their needs shift by occasion, household composition and life stage rather than by traditional demographics alone. Private brands have matured into genuine competitors rather than price-tier fillers. Supply chains have proven fragile in ways we never modeled for. Inflation has fundamentally reshaped value perception, with shoppers trading across tiers, formats and even categories in ways that defy historical elasticity assumptions.”

Consumers are increasingly turning to digital channels to purchase groceries, and AI agents are increasingly part of the baskets being built.

The principles of category management remain as fundamentally sound as when they were first introduced, despite the changes that have taken place in grocery retailing over the decades since. However, Harris says that is not sufficient.

‘Automating a flawed process gives you flawed answers at scale.’

“Executing that process with quarterly reviews and backward-looking scan data simply cannot keep pace with how consumers actually buy today. By the time a traditional category review is presented, the shopper behavior it analyzed has already shifted,” he observes.

Harris says that some retailers are approaching AI with the wrong mindset.

“They’re pointing machine learning at the same scan data and producing the same quarterly decks in less time. That’s automation, not transformation. Automating a flawed process gives you flawed answers at scale. In refrigerated and frozen categories, where shelf life, temperature integrity and velocity assumptions drive everything from assortment to replenishment, that scaled-up flaw shows up as shrink, outof-stocks and missed seasonal windows,” he argues.

New generation category management is distinguished by defining categories around consumer needs and future behavior.

“In frozen meals or refrigerated prepared foods, that’s the difference between catching a shift toward high-protein or single-serve occasions in week two versus quarter three,” Harris explains.

Transforming category practices requires retailers to use AI to answer questions that go beyond what items are sold, the margins they produce and the share the merchant is capturing.

“The new questions are sharper and more consumer- and shopper-grounded. Which shopper missions is this category serving and which is it failing? What unmet needs exist in the household that we’re not addressing on the shelf or online? Which items are truly incremental versus cannibalistic at the trip level? How does this category contribute to retention, not just basket size? In refrigerated and frozen, that often means asking why a shopper bought a single-serve frozen entrée here but their weekly family meals somewhere else — and what we need to change to capture both occasions,” Harris says.

SKILLING UP

The use of AI will provide category managers with tools they’ve never had before, making their jobs easier in many ways while also asking them to develop new skills. They will spend less time on tasks such as assembling data and pulling reports, while focusing more on interpreting consumer opportunities and creating points of difference from their competitors, according to Harris.

“What will distinguish strong category managers is the ability to frame the right questions, challenge model outputs, integrate consumer insight with commercial reality, and lead cross-functional execution. Skills in consumer behavior, experimentation design and supplier collaboration will matter more than spreadsheet proficiency,” he explains.

Harris says grocers that implement AI-supported category management often initially focus on changes to pricing and replenishment because they can get answers to important questions more quickly. Ultimately, though, he thinks the greatest benefits will come from insights around product assortment and shopper engagement.

“Pricing wars can be matched. An assortment built around the missions your shoppers actually have, refreshed continuously, is far harder for a competitor to replicate. This is especially true in refrigerated and frozen, where constrained door space means every facing carries a real opportunity cost and assortment discipline becomes a competitive moat. Promotion is somewhere in between,” says Harris, who adds, “AI improves efficiency, but the bigger prize is rethinking whether the promotion was needed at all.”

The new generation of category management solutions makes the limitations of previous models clear.

“Historical sales tell you what shoppers bought given the assortment you offered. They cannot tell you what shoppers wanted but didn’t find or what they bought elsewhere,” says Harris. “AI allows us to integrate loyalty data, basket composition, digital behavior, external signals and even unstructured feedback to model the household over time — what missions they’re shopping, what life stage they’re in, how their needs are evolving. This moves us from a rearview-mirror analysis to a forward-looking strategy.”

The implementation of AI-powered category management also has benefits that extend to other departments through interconnected workflows.

‘If your category definitions are based on supplier logic rather than shopper missions, AI will optimize within those flawed boundaries.’

“Consumer and demand signals inform merchandising decisions on assortment, price and promotion. Those decisions flow directly into supply chain forecasts and replenishment plans, with constraints fed back in real time. Store operations execute against planograms and availability targets, and the results — sales, out-of-stocks, shrink, shopper response — return to the system to refine the next cycle. The workflow is continuous, transparent across functions, and measured against consumer outcomes, not just departmental KPIs,” says Harris.

Bringing it back to frozen and refrigerated foods, Harris notes, “In refrigerated and frozen, where cold chain integrity, shelf life and tight working capital amplify the cost of any disconnect between functions, this connected loop is not a nice-to-have — it’s the only model that holds up under real operating pressure.”

ONBOARDING AI

Retailers can struggle when using AI to power their category management practices, and it’s not uncommon for challenges to arise because of data quality issues, organizational resistance, process design or simply learning to trust (within limits) the system.

“Data quality issues are real but solvable. Trust in models grows with exposure and good change management. The harder problem is that AI exposes how fragmented and politically encumbered category decisions have become across merchants, supply chain, finance and suppliers. Retailers that try to drop AI onto unchanged processes and incentive structures will not see the return they expect,” contends Harris.

Category management doesn’t just answer what is being bought, but what sales aren’t taking place and why.

AI evangelists strongly believe the technology can help retailers improve customer outcomes. Reaching this point is difficult if grocers prioritize operational efficiency over product relevance, assortment and the shopping experience. Flipping the script means focusing on what consumers value, in addition to profit and loss concerns.

“Track relevance metrics,” instructs Harris. “Are we meeting the missions shoppers come in with? Track availability at the trip level, not just in-stock percentages. Track assortment satisfaction and item findability. Build consumer outcome metrics into the same dashboards used for margin and turn. If efficiency gains come at the cost of shopper experience, you’ve optimized the wrong objective function.”

Harris also sees a need for retailers to determine which types of AI are needed to perform specific functions and to clearly understand the roles of humans and machines in each scenario. A case in point is the use of AI in predictive analytics and optimization models in category management.

“Predictive and optimization models answer ‘What will happen?’ and ‘What should we do?'” distinguishes Harris. These questions are best answered using generative AI (which creates original content in response to a prompt or request).

Harris points to the value of using generative AI to synthesize consumer research, draft individual category strategies, summarize consumer packaged goods supplier proposals and reduce the friction of going from data to decision. “Used well, it removes the human-bandwidth bottleneck that has constrained category management for decades. In refrigerated and frozen, where a single category manager may oversee thousands of SKUs across multiple dayparts and occasions, that bottleneck is acute,” notes Harris.

For all its utility, generative AI is not the be-all, end-all when it comes to category management. Many see agentic AI, which is designed to act autonomously, as the next big step to passive generative AI systems using chatbots. Harris’ firm is among those that have developed agentic solutions.

“An agent doesn’t wait for the category manager to formulate the right question. It monitors signals continuously, identifies when something warrants attention, reasons through the implications, and brings a recommendation forward with its work shown,” explains Harris.

He points out that in frozen novelties, for example, agentic AI “can flag a velocity shift in a competing segment, pull the relevant shopper data, draft a revised assortment recommendation, and surface it to the category manager before the weekly review — not after. That’s the leap from tools that answer questions to a coach and advisor that helps run the category.”

AI’s ability to make good decisions when faced with fast-arriving challenges is particularly important in product categories subject to volatility, seasonality and rapidly changing shopper behavior driven by viral food trends, for example, arising from social media like TikTok.

“AI’s value here is in shorter forecast horizons and more granular signals: store-level, day-level, even daypart-level demand-sensing that incorporates weather, local events and promotional activity. It also helps optimize assortment within constrained door space where every facing has a real opportunity cost. The categories that have historically been managed on instinct stand to gain the most,” adds Harris.

For all its upside, Harris again cautions retailers to avoid garbage-in, garbage-out scenarios when using AI. “If your category definitions are based on supplier logic rather than shopper missions, AI will optimize within those flawed boundaries. If your KPIs reward share growth over consumer loyalty, AI will pursue share at the expense of longterm equity. In refrigerated and frozen, this often shows up as optimizing shrink and turn within a poorly defined assortment — producing efficient execution of the wrong strategy. Worse, the appearance of sophistication can suppress the harder conversation about whether the strategy itself is right. AI amplifies whatever direction you point it in — including the wrong one,” says Harris.

THE HUMAN ELEMENT

Many grocers are focusing their initial AI category management efforts on pricing and
replenishment.

There’s no doubt that AI will increasingly impact the roles of category managers, but the human element will remain central to overall performance.

“Human judgment remains essential for defining the category’s role, setting strategic intent, negotiating with suppliers, navigating brand and reputational considerations, and making trade-offs between short-term performance and long-term consumer equity. These decisions involve incomplete information, competing stakeholders and value judgments that no model can resolve on its own. AI tells you what is likely. Leaders decide what should be,” says Harris. Looking ahead three years, he foresees AI having a profound impact on how category management is practiced within the grocery channel.

“I expect AI to fully handle demand forecasting at granular levels, base price recommendations, replenishment, planogram generation within defined rules, and most routine reporting and insight generation,” says Harris. However, he adds, “Human oversight will remain essential for category role and strategy definition, supplier partnership and negotiation, brand and reputational judgment, ethical trade-offs around pricing and access, and the integration of category strategy with the retailer’s broader positioning. The pattern is consistent: AI handles the bounded and quantifiable. Humans own the judgment-laden and the strategic.”

GROCERS INVEST IN AI

While it’s a challenge to get specifics on AI implementation by grocery retailers, particularly in category management, there have been enough public pronouncements to see where some of the biggest players in the industry are focusing their efforts.

Albertsons Companies has been the most explicit in tying AI to merchandising and category management. Anuj Dhanda, chief technology and transformation officer at Albertsons, cites merchandising intelligence as one of the four big bets — digital customer experience, merchandising intelligence, empowered associates and leaders, and supply chain optimization — it is making in AI.

“We believe AI-driven insights have the potential to transform category management. By equipping merchants with tools to optimize promotions, assortment and the decisions they make weekly, we will move from reactive to predictive merchandising intelligence for our teams. This is about precision and accelerating decision cycles in a highly competitive environment using data to anticipate our customers’ needs and deliver value for them more effectively and quickly,” writes Dhanda.

Kroger is also stepping up its AI efforts, as evidenced by the company’s creation of a chief data and AI officer role this year. Milen Mahadevan, president of Kroger’s 84.51° data science subsidiary, will lead this effort and report to Yael Cossett, executive vp and chief digital and technology officer.

The supermarket giant has expanded its partnership with Google Cloud to use Gemini Enterprise, a platform aiding consumer grocery planning via integrated meal and shopping assistants. For Kroger category managers, that means that AI will influence replenishment, pricing and how shoppers are introduced to products and choose which ones to buy.

From its public disclosures, Walmart seems focused on using AI to improve productivity and workflow, but like Kroger, the steps it is taking appear likely to influence assortment, planning and execution. The company unveiled plans last year to launch four “super agents” that would make it easier for customers to shop and for the retailer’s teams to streamline operations.

Walmart’s Sparky shopping assistant is at the center of its efforts to reshape how customers interact with its product selection, and early results are encouraging. CEO John Furner said that an analysis of Walmart’s Q4 2026 earnings showed that Sparky “drives 35% higher average order values compared to non-Sparky sessions” and that half of all Walmart app users have used Sparky to shop. The decisions that Walmart’s Sparky-aided shoppers make will affect how the chain’s category managers do their jobs.