Procter & Gamble cuts model deployment time up to 90% with Azure IoT Operations

Procter & Gamble (P&G) has a nearly 200-year history of irresistible product superiority—a key reason why 99% of US households use at least one P&G product to care for their health, homes, and loved ones. 

Earning and maintaining such deep customer loyalty for nearly two centuries requires a high degree of agility across the organization. P&G’s agility is reflected in the company’s commitment to digital acumen—tapping into relevant data and technologies to delight consumers and retailers while driving innovation, streamlining the supply chain, improving quality and decision making, and increasing productivity.

Of course, maintaining industry leadership also requires each business unit to harmoniously own its use cases. And in manufacturing, that means ensuring details like the exact sheet length of a Bounty paper towel or applying the perfect amount of glue on a Pampers diaper. Going one step further, P&G must do this consistently across plants in more than 35 countries, using a mix of new and aging production equipment.

Meeting that challenge was a key reason the company began a multiyear collaboration with Microsoft in 2022. P&G manufacturing and technology leaders knew that by working together to contextualize data and apply analytics, the company would create a common data plane to improve consistency and overall equipment efficiency (OEE). 

Aligning IT and OT across production lines

With manufacturing facilities in many different areas, P&G’s equipment can vary based on country-specific manufacturing availability, which can make standardization a challenge.

“Where IT meets OT, we were not as standardized as we wanted to be. Building AI models which are executed on the OT—what we called industrial IoT—was not as easy as we hoped,” explains Kurt Lermann, Vice President of Information Technology at P&G. “While we delivered on pilot after pilot, all the variations in our equipment made things difficult when we went to scale.” Looking for a way to accelerate manufacturing transformation and knowing Microsoft’s commitment to advancing industrial IoT, P&G chose the Microsoft Azure IoT Operations platform built on top of Azure Arc, as the solution.

Using Azure IoT Operations, P&G was able to factor in differences in equipment and remove barriers between operational technology (OT) and IT systems, making it possible for toothpaste lines in plants running disparate equipment to produce an identical product. The solution brings together a combination of hybrid cloud infrastructure, AI, and enhanced workload orchestration to seamlessly deploy cloud-native, Kubernetes-based applications directly to P&G’s global manufacturing footprint. The company uses Azure Arc to build and orchestrate Kubernetes-based models in Azure in the cloud and deploy them at the edge on equipment in P&G manufacturing facilities, streamlining the entire DevOps pipeline.

“Having everything contained and deployable on the most suitable runtime environment is a key initiative I’m driving within the organization. Azure Arc plays a key role in helping us develop in the cloud and deploy on the edge,” explains Greg Hormann, Senior Director of Information Technology at Procter & Gamble.

Enhancing manufacturing agility       

A Kubernetes-native MQTT broker provides the messaging plane for Azure IoT Operations, enabling bidirectional edge-to-cloud communications and supporting event-driven applications at the edge. “The Azure IoT Operations message broker makes sure every application gets the right information as it’s coming in from the manufacturing equipment and pushes the information to where it wants to go,” Hormann says.

P&G can now capture data—such as temperature or the length of a paper towel—and extrapolate insights from those physical devices at the edge. That means the company can ensure greater reliability by monitoring and anticipating anomalies, helping avoid unplanned downtime on equipment. 

“Microsoft technology helps us take advantage of analytics to determine and ensure everything is working exactly as it should be, and if it isn’t, trigger a more detailed inspection by one of our operators on the line,” adds Hormann. “That allows us to avoid potential quality issues and keep operators focused on important work.”

The Azure Arc workload orchestration feature facilitates deployment at scale and allows operators to make adjustments on the floor. Using workload orchestration, P&G can easily deploy a new version of a model. The company can also scale multiple models across sites with different configurations to produce virtually identical results with minimal operator effort. When P&G needs to update a model on the line, it happens fast with the company experiencing as much as a 90% reduction in the time required to deploy a new version.

“We're really excited about workload orchestration,” Hormann says. “We run a lot of workloads but haven’t had a good interface that allows people on the lines to configure applications and change a tolerance setting. Workload orchestration gives operators the ability not only to make small tweaks to their application and control it themselves but also control the scheduling of an upgrade from one version of an application to another on their line—and time it exactly, avoiding any unexpected downtime.”

Enabling AI at the edge 

By bringing together a hybrid cloud infrastructure, AI, and workload orchestration, Azure IoT Operations allows P&G to configure endpoints, data flows, and edge assets to capture edge data and feed its AI models. Using that data and AI, models can be retrained and redeployed directly at the edge. 

“With Azure IoT Operations, we’re able to run near real-time machine learning models at the edge that do predictions or quality monitoring and move quite a bit of data to our core data lake hosted on Azure. That allows us to perform analytics, look for patterns, and uncover new opportunities to improve our processes and products,” Hormann explains.

“Data from the line travels edge-to-cloud into P&G’s Azure-based corporate data lake. P&G data scientists use that data to build AI models, look into history, detect trends, and drive predictive or prescriptive models,” says Lermann.

With an AI-enhanced ability to deploy services wherever they’re needed, manage edge resources from the cloud, and bring workloads into a unified environment, IT and manufacturing teams at P&G can spend more of their time on strategic initiatives. This shift helps the company deliver consistent quality and innovation and delight consumers through irresistible superiority across its product portfolio.

Reflecting on the new platform’s impact at P&G, Lermann shares, “To me, this technology is like moving to the cloud years ago—a prerequisite to doing a lot of the things we’ve been able to do since. And it’s the same with our Azure IoT Operations platform. It will enable us to stay relevant.”

Discover more about Procter & Gamble on Instagram, LinkedIn, and YouTube.