How retailers should split workloads between cloud and edge for real-time store operations

What should run in the cloud, and what should run in the store? Match each workload to the environment it was built for. Real-time work belongs at the edge. That means POS transactions, shrink detection, video recognition and in-store promotions. These process locally and keep running even when connectivity drops. Use the cloud for the rest: offline data analysis, AI model training, big-data processing and eCommerce. So the answer is not “Cloud Out” or “Edge In.” It is both, built as one hybrid architecture.

Key takeaways

• Real-time store workloads such as POS, payments, video AI and kitchen systems belong at the edge because latency and connectivity failures make cloud-only processing unreliable in distributed store environments.
• Cloud infrastructure excels at offline analytics, AI/ML model training, big-data processing and eCommerce. These jobs need concentrated computing power, not fast local response.
• Cloud POS is really a hybrid solution. It must be edge-enabled to keep working when connectivity drops. That makes it one of the biggest use cases for edge computing.
• Retailers should avoid repurposing cloud infrastructure for the edge. Edge requires purpose-built tools designed for tens of thousands of small, poorly connected locations running 24/7.

Data volumes are growing. Stores are getting more connected. Success depends on putting the right workloads in the right place. A hybrid cloud-edge approach enables real-time processing at the store. At the same time, it unlocks long-term value through centralized analytics.

Edge computing in retail is growing fast. More devices are connected in stores. More applications depend on the data they create. The role of clouds and corporate data centers is clear too. They crunch transactions and provide data insights. Retail IT teams have spent a lot of time on their overall cloud strategy.

Now their focus is turning to the edge. Stores process large amounts of data there, and not just transactions. Think inventory, sensor, capacity, video and performance data across tens to hundreds of devices in each store. That shift is already reshaping grocery IT priorities. Nearly 70% of US grocery IT leaders expect most in-store software to run as cloud-native microservices in containers within one to five years. In Europe, 57% say the same. That comes from NCR Voyix research conducted with Censuswide.

Store infrastructure directly shapes the customer experience. It is no longer a back-end concern. That creates both a challenge and an opportunity. How do you optimize edge systems and the data they create? And how do you turn that into a competitive advantage?

Edge technology is the set of systems that operate outside the cloud and corporate data center. It is part of a bigger retail architecture. That architecture includes public cloud, SaaS solutions, eCommerce stacks, distribution centers and customer-owned or leased data centers. All of it operates “above store” and runs across the network to the front-line systems.

In retail, a cloud strategy without an edge strategy is incomplete. At worst, it will fall short.

Many retailers see edge technology as the best way to modernize their store infrastructure. Everyone agrees that retail success needs a blend of cloud and edge. But should you take a “Cloud Out” or “Edge In” approach?

I'll help you decide.

Why cloud and edge serve different jobs in retail

The cloud has transformed how applications run and broken barriers to innovation. Several things drove that: virtualization and containers, better broadband, and heavy investment by hyperscalers like Google, Microsoft and Amazon. But like any tool, cloud works well for some jobs and not others.

Edge is built for conditions that cloud infrastructure was never meant to handle. Unlike data centers, edge infrastructure manages unreliable networking. Latency, jitter and availability issues are common, especially across large distributed store estates. Store systems rarely sit in people-free, highly secure, air-conditioned server rooms. Quite the opposite.

Edge systems have a lot to handle. They process and control high volumes of checkout and payment data. They run weigh scales and kitchen systems. They manage camera feeds with AI/ML working in real time. Above all, they deliver high performance at peak hours, often running 24/7.

Infrastructure built for the cloud is a poor fit for the edge. The reverse is also true. Cloud systems usually run a handful of large, universally available locations. People call this “operating at cloud scale.” Edge systems are different. They run as many as tens of thousands of small, often poorly connected locations. That is “operating at edge scale.”

At a high level, both need scalability and automation. Under the covers, one is highly concentrated and the other is highly distributed. Retail success comes from designing systems that run continuously across both. Using both is the foundation of unified commerce. It delivers consistent, real-time experiences across every store and channel.

How cloud and edge work together for real-time retail operations

Retail needs real-time processing. Think video recognition at frictionless stores, shrink detection, product scanning, or the latest personalized in-store offers. Latency and high bandwidth costs create friction. That hurts both the customer experience and the bottom line. Edge systems process this data in real time. But analyzing it over a longer period takes serious offline computing power from the cloud.

The decision heuristic: "Edge In" for stores, "Cloud Out" for above-store

A distributed store estate has many moving parts. So edge infrastructure must be responsive, scalable and reliable. “Edge In” is the best approach for stores. Systems in headquarters, above-store data centers or eCommerce are more cloud-like. They fit a “Cloud Out” strategy better.

Why cloud versus edge is not a binary choice

Here is the reality. You design for the store first, at the edge. Then you make the most of the cloud. Cloud is great for offline data analysis, building learning models, processing big data, eCommerce and more. It is central to your design, but not on its own.

Say you want real-time promotions delivered to shoppers in your store. You host the application at the edge for instant responses. Then you train the artificial intelligence (AI) algorithm in the cloud and push model changes back out to the edge.

We also need to be clear about the word cloud. Take cloud point-of-sale (POS). POS is one of the biggest use cases for edge computing. It processes millions of transactions. But there are many potential failure points: switches, routers, networks and cloud providers. Imagine POS as a true cloud application, a web browser connected over the internet to a cloud. If connectivity dropped, there would be no POS in the store. So cloud POS has to be edge-enabled. That makes it a hybrid, both cloud and edge.

How a hybrid approach simplifies operations and reduces cost

Cloud Out and Edge In are not mutually exclusive. To get the best of both, you need a holistic approach. Only then can you capture the cost efficiencies. The opportunity is big. In NCR Voyix research with Censuswide, 93% of US grocers said they do not yet have a fully implemented retail edge plan. In Europe, the figure is 85%. That leaves real efficiencies on the table.

Why stream all data to the cloud 24/7? A gateway function lets you choose which data goes to the cloud and which stays at the edge. A hybrid approach simplifies operations.

Why edge infrastructure can't be repurposed from the cloud

Edge computing does not compete with cloud computing. They complement and complete each other. In a store, edge is in some ways just a distributed cloud. But it is a very different type of cloud.

Cloud computing has delivered efficient resources and centralized applications on a pay-as-you-go basis. Hyperscalers are slowly working out how to push the most relevant parts of their technology out of those data centers and into other clouds and data centers.

But beware of infrastructure that has been repurposed from cloud to edge. You would not ask the team that built a large nuclear power station to use the same approach for thousands of wind turbines hundreds of miles offshore. Edge needs tools designed for a different job.

Edge is a chance to consolidate old infrastructure and drive efficiency. It also opens up new high-data, low-latency use cases that improve the customer experience. The key is to integrate your cloud and edge. Do that, and you can deploy and manage applications where they belong, without long development cycles. You also get consistent policies and configuration across the whole architecture.

Focus on business outcomes, not technology problems

As retailers make this transition, the focus should be on business benefits, not on solving specific technology problems. This approach lets you quantify outcome-based solutions with specific service level agreements (SLAs). The financial case is measurable. A Total Economic Impact study by Forrester Consulting, commissioned by NCR Voyix, found strong results for Voyix Edge: a 164% return on investment, a payback period of six to 18 months, a 50% reduction in unplanned downtime, and a 10% improvement in store performance and efficiency. Outcomes like these are what make outcome-based SLAs credible.

So it is not “Cloud Out” versus “Edge In.” It is how you combine the two to get the most from your in-store operations and drive success. Follow these best practices, work with the right providers, and you can find the right answer for your business.

Retailers want the same agility in the store that they already have in their digital channels. Edge is a key technology to help them get there.

Nick East is Chief Product Officer at NCR Voyix, where he leads product strategy across the retail and restaurant technology portfolio. With more than 25 years driving software innovation, he was co-founder and CEO of Zynstra, an award-winning leader in purpose-built retail edge software acquired by NCR Voyix in 2019, and has since been central to developing the company's edge-enabled commerce platform.

Frequently asked questions

What is the difference between cloud scale and edge scale in retail?

Cloud scale means running a handful of large, universally available data center locations with concentrated computing power. Edge scale means running tens of thousands of small, often poorly connected store locations with distributed processing. Both need scalability and automation. The difference is simple: cloud is highly concentrated, and edge is highly distributed.

Why does cloud POS need edge computing to work in stores?

Imagine POS as a true cloud application, a web browser connected over the internet to a cloud. If connectivity dropped, there would be no POS in the store. Cloud POS has to be edge-enabled so transactions keep processing locally during network outages. That makes it a hybrid, both cloud and edge.

Can retailers repurpose cloud infrastructure for edge computing in stores?

Generally not. Infrastructure built for the cloud is a poor fit for the edge. Edge environments face unreliable networking, uncontrolled physical conditions, and the need for 24/7 real-time processing across thousands of locations. Use purpose-built, edge-native infrastructure for this distributed environment. Do not force cloud tools into the store.

The right technology helps your business move with your shoppers’ expectations. But to reach even greater heights, you need seamless experiences. To learn more, read our article on store transformation with experience-driven retail.

‍

Sources

1. NCR Voyix research conducted with Censuswide, "Grocery Store IT: The Edge Revolution," survey of senior grocery IT decision makers in the US and Europe. Cited for the containerized microservices adoption figures (70% US / 57% Europe) and the retail edge plan readiness figures (93% US / 85% Europe). Read the Article
2. The Total Economic Impact of NCR Software Defined Store (now Voyix Edge), a commissioned study conducted by Forrester Consulting on behalf of NCR. Cited for 164% ROI, payback of six to 18 months, 50% reduction in unplanned downtime, and 10% improvement in store performance and efficiency. Read the Forrester Report