SALESFORCE

How to build an omnichannel retail business ready for agentic AI with Salesforce

Publication date: November 14, 2026

Author: Manuel García Gonzalo, Director of Salesforce Strategy and Business Development at Inetum

For years, omnichannel retail has been one of the industry’s biggest goals. Organizations have invested in e-commerce, mobile apps, loyalty programs, marketplaces, and new service channels to create more seamless customer experiences. Yet many of those initiatives have failed to solve one of retail’s most persistent problems: fragmented customer experience.

The real problem was never the channels

Omnichannel is often approached as a technology project centered on adding new sales channels. That leaves the more fundamental challenge unresolved: how information is managed and shared.

A customer remains the same person whether they buy online, check a promotion in a mobile app, or walk into a physical store. Yet many organizations still maintain separate profiles for those interactions. The result is predictable: irrelevant recommendations, duplicate promotions, inventory problems, and an inconsistent experience.

Customer expectations have changed at the same time. Making a purchase convenient is no longer enough. Consumers increasingly expect brands to understand their preferences, anticipate their needs, and maintain a consistent relationship across channels. Their benchmark is not necessarily another retailer. It is the best digital experience they have had anywhere.

Meanwhile, margin pressure is forcing retailers to do more with fewer resources. Personalizing every interaction through human intervention alone is difficult to scale. Intelligent automation is therefore becoming a competitive necessity.

The issue is not the number of channels. It is the data behind them and the processes that put that data to work. When the same customer is recognized differently in a store, on a website, and in a mobile app; when inventory numbers do not match across systems; or when marketing, customer service, and operations work from different information, omnichannel becomes little more than a collection of disconnected channels.

The next phase of omnichannel retail is about solving that problem. Retailers do not need more touchpoints for their own sake. They need to connect every channel around a unified view of the customer and design processes that artificial intelligence can act on in real time. That is the real foundation for sustainable omnichannel growth.

Data is where omnichannel retail really starts

The difference between a traditional omnichannel strategy and a model built to scale is where the transformation begins.

For years, companies deployed channels first and tried to connect the information afterward. That produced complex, costly environments that were difficult to scale.

At Inetum, we propose reversing that logic through a reference model that structures the core areas of retail transformation. Data sits at the foundation, with a unified customer view and strong information governance. Channels, operations, and AI capabilities are then integrated on top of that common foundation.

This approach allows retailers to move in phases and generate measurable results early, instead of concentrating risk, budget, and implementation effort in a large transformation before value has been demonstrated. The goal is not simply to deploy more technology. It is to create a common foundation that lets the entire organization operate from the same information.

Salesforce Data 360 as the foundation for Customer 360 in retail

Creating a unified view requires a platform that can connect customers, operations, and channels in real time.

In this model, Salesforce Data 360, until recently known as Data Cloud, serves as the backbone by consolidating customer, product, inventory, and digital behavior data into a single, continuously updated profile.

Commerce Cloud, Marketing Cloud, Service Cloud, and Agentforce agents operate on this common data foundation. Marketing can personalize campaigns using the same customer knowledge that service teams use to resolve an issue or a store uses to confirm whether a product is available.

The result is a more consistent experience across touchpoints and an organization that can operate from a common source of information.

An architecture built for any interaction

Delivering truly omnichannel experiences also requires an architectural shift.

For years, many e-commerce environments evolved into platforms where each channel maintained its own business logic. That made innovation harder and created another layer of technology silos.

Headless or composable architecture changes the model by separating the user experience from core business capabilities. Inventory, promotions, orders, catalogs, and customer profiles can then serve multiple channels without continually rebuilding the back end.

Headless 360, the evolution of this model that Salesforce introduced in 2026, takes that approach a step further. It exposes platform business capabilities so that both people and intelligent agents, whether Salesforce-based or third-party, can interact directly with Salesforce processes and data while respecting established permissions and governance rules. That can accelerate the deployment of new experiences while reducing integration complexity.

From a sales channel to an intelligent interaction

Once systems operate from the same information, channels stop being simple transaction points. They become spaces for continuous interaction among the brand, the customer, and employees.

This is where one of the most important shifts in retail emerges: agentic AI.

Unlike traditional virtual assistants, AI agents can understand context, make decisions, and act on business processes in real time.

Within Salesforce, Agentforce provides the agent layer of the platform, connecting agents to the same Data 360 information used by other applications. Those agents can work with customer identity, purchase history, previous interactions, and current inventory information.

That enables use cases such as recommending alternatives when an item is out of stock, managing returns, or anticipating replenishment needs. The interaction moves from reactive service toward proactive assistance designed to create value.

Without actionable processes, useful AI agents are impossible

There is a step that is often overlooked in conversations about agentic AI: unified data is necessary, but it is not enough.

Most retail processes were designed for people to execute sequentially, often relying on implicit judgment. A return may depend on the store manager’s decision. A promotion may be applied “depending on the circumstances.” A customer issue might be resolved over the phone without leaving a structured record.

An AI agent cannot reliably act on that kind of process, no matter how good the underlying data is.

That is why much of the work in the projects we are tackling involves redesigning processes, so they are actionable by AI. Each step needs explicit data, defined decision rules, and a concrete action that an agent can execute or escalate to a person.

This work is less visible than launching a conversational assistant, but it is what separates an agent that answers questions from one that resolves them. In my experience, this is where retailers should focus before deciding which agents they want to deploy.

Loyalty is built through utility

For decades, loyalty programs have centered on discounts, promotions, and points. Those mechanisms still matter, but it is increasingly difficult to differentiate on financial incentives alone.

The stronger competitive advantage comes from utility.

When a grocery retailer remembers shopping habits, suggests personalized recipes, alerts a customer when frequently purchased products are available again, or adapts recommendations to specific needs, the relationship changes.

Instead of receiving mass communications, customers get relevant, contextual recommendations. Loyalty becomes less dependent on price alone and more closely tied to experiences that make the customer’s life easier.

From data to results: a real-world example

The potential of this model can already be seen in projects developed by Inetum.

At Barceló Hotel Group, the company has introduced agentic AI on Salesforce to transform the management of commercial inquiries.

When a customer requests a quote, an AI agent automatically analyzes the message, identifies intent, classifies the inquiry, and extracts relevant information such as dates, the requested property, and the number of guests. If information is missing, it continues the interaction until the required details have been collected.

By the time the request reaches a human team member, the information is already structured and ready for preparing a proposal. Simpler inquiries can even be resolved automatically.

Beyond saving time, interactions are recorded against a unified customer profile. This makes it possible to measure the full journey, identify opportunities for improvement, and continually enrich the customer knowledge available to the organization.

The same approach applies directly to retail, with the additional advantage that many sales and after-sales processes are already more standardized than group hotel sales. An intelligent agent can assist during a purchase, provide availability information, recommend alternatives, or manage customer issues using the same information available to physical stores, e-commerce platforms, and service teams.

Grow through efficiency, not complexity

With retail margins under increasing pressure, sustainable growth depends on improving customer experience and operational efficiency at the same time.

Unified information can help optimize inventory and reduce stockouts. Agentic AI can lower service costs, while personalization can improve conversion and average order value.

According to the IHL Group study on inventory distortion referenced in the original article, annual losses caused by stockouts and excess inventory approach $1.77 trillion globally. An integrated view of inventory combined with predictive capabilities is therefore more than a technology consideration. It can become a direct lever for profitability.

Omnichannel transformation, however, should not be treated as one massive deployment. Experience shows that it is more effective to establish a strong data foundation, identify high-impact use cases, and expand progressively as each new capability produces results.

That is the approach we advocate at Inetum: build a genuine Customer 360, establish a common information foundation, and use it to support omnichannel experiences and intelligent agents that can evolve with the business.

I have spent years watching retailers add channels faster than their organizations can absorb them. My conclusion is simple: continuing to add touchpoints on top of fragmented data and processes designed exclusively for people will not take retailers where they need to go.

The work that matters now is less visible. Unify the data and redesign processes so agentic AI can execute them. That is what turns AI agents into business outcomes, and it is where retail should focus next.