SAP Product Chief Says Data Foundations, Not Storefronts, Now Define Retail Success

Agentic commerce has moved quickly from a future-facing concept to an operational reality, as software agents increasingly mediate how consumers discover, compare and buy products. What was once framed as experimentation is now becoming a competitive requirement, forcing enterprise retailers to confront whether their data, systems and customer experiences are built for an agent-driven economy.

That urgency was a central theme in a recent Monday Conversation between Karen Webster, CEO of PYMNTS, and Balaji Balasubramanian, president of SAP Customer Experience. Their discussion focused on what agentic commerce means in practice, and why many large retailers are still underprepared.

Enterprise Retail Scores a Three or Four

When asked to assess enterprise retail’s readiness for agentic commerce on a scale of 1 to 10, Balasubramanian did not hesitate. “Certainly it’s not 10,” he told Webster. “It’s far from it.” He later added that, in his view, most enterprise retailers sit “probably at a 3 or a 4 at best.”

The issue is not a lack of data. Rather, it is how that data is organized, harmonized and accessed. Balasubramanian pointed to years of system sprawl that has left information siloed across platforms, formats and business units. Even when the data exists, it is often not assembled in a way that can support real-time decision-making by either humans or machines.

The debate over whether agentic commerce will matter has effectively ended. According to Balasubramanian, retailers are no longer asking if the shift is happening. They are asking how to respond to it.

Retailers, he explained, see evidence daily that discovery is moving upstream into answer engines and AI-driven interfaces, often without brand preference built in. If retailers are not discoverable in those environments, they risk being invisible altogether.

Bots Are Outpacing Humans

One of the more striking signals of change is traffic itself. Balasubramanian noted that bot traffic has already exceeded human traffic, a trend that retailers are watching closely. At the same time, traditional organic search is weakening as top-of-funnel discovery collapses and shifts into AI-mediated environments.

Those agents, whether acting autonomously or assisting humans, require dense, structured and current data. As Balasubramanian explained, the “agent web” is fundamentally different from the human web, demanding richer metadata to interpret intent and return actionable outcomes.

Across the discussion, data emerged as the foundation upon which agentic commerce is built. Retailers with the strongest data foundation will be best positioned to embrace and harness agentic commerce, Balasubramanian stated.

That foundation must support both structured and unstructured data, and it must be kept near real time. Without it, agents cannot surface products accurately, availability cannot be confirmed, and offers cannot be personalized at scale. The result is missed discovery and lost demand before a retailer even has a chance to compete.

“What you need on an agent web is denser metadata … to feed the channels and the behaviors that we are seeing,” he said.

Siloed Systems, Fragmented Reality

Webster underscored that the data challenge lies not just volume, but accessibility. Much of retailers’ data, she observed, is “not assembled in a way to be actionable” and “not assembled in a way that can be accessed in real time.”

Balasubramanian agreed, noting that no two retailers are alike. Enterprise retailers operate across different subverticals, channels, geographies and inventory models. That complexity makes harmonization difficult, but also unavoidable if agentic commerce is to work consistently across touchpoints.

As president of SAP CX, Balasubramanian hears the same questions repeatedly from clients. Retailers want to know how to prepare for agent-driven discovery, how to ensure their products surface in AI responses, and how to connect customer interactions to planning and fulfillment.

SAP’s answer, he said, lies in building an end-to-end data foundation that unifies product information, availability, inventory, supply chain and pricing into a single, coherent view that can serve both humans and agents.

“We talk to them about loyalty at a different level,” he told Webster, “which is experiential loyalty — the type of experience that you serve to customers that will help them be much more affiliated to a brand and convert them into long-lasting fans.”

Types of Agentic Commerce

Effective data utilization must distinguish between different modes of agentic commerce. Some models involve agents executing transactions end to end based on a prompt. Others keep humans in the loop, with agents assisting discovery and decision-making. In both cases, success depends on whether retailers’ products and offers are visible and interpretable to machines.

If products do not show up with sufficient context, the experience becomes inconsistent and fragmented, undermining both conversion and loyalty, said the SAP executive.

Balasubramanian outlined three priorities for retailers building toward optimal agentic commerce. Product content includes attributes and metadata that support discovery.  Availability covers inventory and fulfillment. Offers bridge awareness and conversion through relevance and timing.

Together, those elements allow agents to move seamlessly from intent to action, whether the shopper is human or automated.

Loyalty Beyond Points

Traditional earn-and-burn loyalty programs still have a role, Balasubramanian said, but they are no longer sufficient. SAP’s focus is on experiential loyalty, built through personalized interactions, flexible purchasing options and post-purchase engagement.

By connecting discovery, transaction and planning in near real time, retailers can forecast demand more accurately and deliver stronger offers. That, Balasubramanian argued, creates repeat buyers and long-term brand advocates.

Agentic commerce will further redefine how retailers think about customer relationships. As Webster observed, the challenge is about how brands become part of the prompt itself, rather than relying on advertising or influence.

For Balasubramanian, the path forward is clear. “Data foundation is how you can win in the ‘agent era,’” he said.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.