Verizon created an integrated supply chain planning platform for its consumer mobile phones and accessories business, unifying previously fragmented business processes to provide a single source of truth and using artificial intelligence (AI) and machine learning (ML) to enable proactive decisions to mitigate supply disruption, optimize inventory, and save millions of dollars in capital spend. How: Verizon’s OnePlanning program uses AI and machine learning to raise its supply chain system’s accuracy.
Verizon uses AI and machine learning to optimize supply chain inventory
Who: Verizon Communications Inc.
What: Created an integrated supply chain planning platform for its consumer mobile phones and accessories business. Called OnePlanning, the platform unifies previously fragmented business processes to provide a single source of truth and uses artificial intelligence (AI) and machine learning (ML) to enable proactive decisions allowing Verizon to mitigate supply disruption, optimize inventory, and save millions of dollars in capital spend without compromising customer service or partner relationships.
How: Verizon’s OnePlanning program uses AI and machine learning to raise its supply chain system’s accuracy. It incorporates advanced statistical models, predictive analytics and end-to-end automation to calculate inventory targets, optimize inventory and deliver faster, more accurate decision making. TM Forum’s Business Process Framework (eTOM) and Open APIs provided common vernacular, flexibility and extendibility to business process management towards enterprise wide standards, simplified business processes, and interoperability among partners.
Results: One Planning has improved supply chain data integrity and accuracy; provided direct users with a single source of truth; made it easier to test out real-world scenarios and adjust plans, provided end-to-end supply chain visibility and improved collaboration with partners, including suppliers and OEMs. It has improved customer experience by providing just in time delivery to retail locations across the US. As part of the rollout, the forecast accuracy for mobile phones have improved by a staggering double digits, leading to lower inventory levels and more optimal buying and replenishment at retail stores. In addition, expedited delivery to retail stores have reduced significantly, thereby saving transportation costs.
Every large retail organization knows that the more visibility, intelligence and agility a supply chain management system provides, the better equipped it is to cope with ebbs and flows in customer demand, as well as unforeseen shocks to product supply. So, when Verizon set about overhauling its supply chain unit for mobile phones and accessories it used AI and machine learning to create a smarter, integrated, more automated system that could anticipate supply chain disruptions more quickly and allow Verizon to optimize inventory and customer experience.
Each year, Verizon sells billions of dollars of mobile phone and accessory assets and can have a significant inventory in stock at any one time. How well it manages this inventory has a huge impact on both customer service and its finances. “We have to balance between carrying a robust inventory of devices for our customers to choose from, while also optimizing our capital spend and inventory in our stores and warehouses,“ explains Jhuma Nath, Director, Corporate Systems Group, Verizon. Prior to deploying OnePlanning many of Verizon’s supply chain processes were manual and transactional and plans often relied on a large amount of manual assumption and computation, making it difficult to quickly create accurate forecasts, especially given the size of Verizon’s business.
In addition to ensuring Verizon can stock what customers want, the new supply chain had to provide the necessary intelligence and agility to maximize operational cost efficiency. “We had to reduce working capital, transportation costs and warehouse costs without compromising on-time availability, service levels and customer satisfaction,” explains Nath. Covid-19, meanwhile, illustrated the importance of creating a resilient, agile system that can deal with the unforeseeable.
Before the introduction of OnePlanning, Verizon’s supply chain was managed via an array of business processes and tools. Business teams had to grapple with decentralized data sources to gain a complete picture of available inventory, spending more time on handling data.
They also lacked the right tools for simulation and scenario planning, leaving decision-making up to individual planners, who each possessed different knowledge, skills and experience. The OnePlanning program therefore set out to remove silos and create an integrated system that gave all team members a single source of truth based on accurate data.
Another disadvantage of a siloed supply chain operation was inconsistent collaboration with supplier and third-party logistics companies. Communication went through an ad-hoc mix of channels, including e-mail, which increased the risk of errors or delays. In addition, the previous system took a macro level approach to planning, and any adjustments were tedious and time-consuming to make.
Verizon took a multi-pronged approach to building an end-to-end, automated supply chain system that gives planners real-time visibility of accurate data and allows them to experiment with different scenarios and adjust outcomes. They transformed and automated business processes across the board, including the key areas of product lifecycle management, inventory planning and forecasting, and supplier integration. Before introducing the OnePlanning program, for example, Verizon’s product lifecycle management teams had to manually mark a product’s end of life within the system. Not only was the task laborious, it did not allow for nuanced forecasting. Verizon therefore equipped its OnePlanning program with advanced planning algorithms that are able to take into account the ramp-up and ramp-down of phases within a product’s life cycle, emulating and learning from similar products used in the ecosystem.
OnePlanning has also dramatically altered Verizon’s ability to perform scenario planning. Planners can now tap into simulations of real-world scenarios and see in real-time how changes would impact demand, supply or inventory and then use the information to make decisions and when dealing with partners. Meanwhile, new business processes have improved communication with Verizon’s numerous suppliers and OEMs. Whereas before purchasing plan agreements and updates were tracked through emails and meetings, OnePlanning enables Verizon and its suppliers to simultaneously share and view the same data through a portal.
In addition to enabling automation, the new system draws advanced statistical modeling, predictive analytics and machine learning algorithms to drive greater accuracy and improve the intelligence that planners have at their disposal. Planners can now deploy best fit forecasting models instead of less precise moving averages. OnePlanning uses Mean Absolute Percentage Error (MAPE) to select the best fit forecast models based on historical information and predictive analytics. As a result, Verizon can apply seasonal changes – such as the launch of a smartphone by a major brand - and promotions to accurate baseline forecasts and ensure their plans adapt accordingly. Verizon also uses the OnePlanning platform to more accurately forecast and manage product availability in its retail outlets on a store-by-store basis.
This is done with an intelligent store allocation and replenishment platform, which takes into account inventory demands across multiple locations and ordering to create Unified Demand Forecasts (UDF) as well as new product launches, promotions, seasonal variations and historic patterns, otherwise known as Demand Influencing Factors (DIF). Its machine learning libraries then apply predictive analytics to UDF and DIF to generate complex demand forecasts, replacing manual calculations that were previously performed in functional silos. In addition, Verizon uses AI and ML to improve its safety stock calculations. Safety stock is a buffer that retailers maintain to mitigate against the risk of unforeseen drops in supply. OnePlanning makes it possible to optimize safety inventory by applying AI and ML to retail and distribution center data.
OnePlanning has driven automation and transparency with variety of suppliers and vendors towards real-time and agile decision making. Verizon adopts industry best practices, tools and integration frameworks that enables easier and faster supplier onboarding. 50+ suppliers have been integrated within this ecosystem in 6-9 months, powered by modular, reusable building blocks. We have leveraged design principles and best practices of TM Forum’s Business Process Framework (eTOM) towards supply chain strategy and planning, supply chain capacity delivery, product lifecycle management, supplier/partner relationship management and interoperability among partners.
The flexible, extendable framework and standards based common vernacular allows us to create business processes that meet Verizon specific needs while driving speed of integration. Business benefits go far beyond customer experience. This transformational program has now positioned Verizon to optimize overall working capital by assessing the availability of inventory, calculating supplier affordability, and determining which stores should receive specific devices and accessories. It also enables optimization of business processes such as device shipment to reduce transportation costs. Importantly OnePlanning equips Verizon to evolve its supply chain in line with future needs, says Nath. “The way the platform is built there are lots of hooks in place if we need to put in advanced algorithms and machine learning, positioning us for the next five to ten years.”