Coca-Cola Beverages Africa (CCBA) is the eighth largest authorised bottler in the world by revenue, and accounts for over 40% of all Coca-Cola drinks sold on the continent by volume. Barry Anderson is group planning leader at the company and it’s his job to ensure the right products and materials are available at the right time, from raw materials and packaging to the finished drinks. This matters because a stock out is more than a missed sale; it means lost revenue, frustrated customers, and a competitor gaining ground.
In the past, ensuring stock was on the shelves involved forecasting, spreadsheets and a lot of firefighting, says Anderson. “At its simplest, a planner might look at the previous year’s numbers and use them to estimate what demand will look like this year. But this is very inefficient. Every Monday, we were breaking down the old plan and rebuilding it from scratch because something wasn’t working.”
CCBA operates in 14 territories, and has 35 bottling plants and more than 840 000 customer outlets. To put the scale of the business in context, in 2025 alone, the company shipped around 1bn unit cases. “Changes in customer preferences mean that we’re dealing with more brands, more flavours, and more variability in product packs, such as a single 500ml bottle versus a six-pack of 500ml.” He says this makes effective forecasting very difficult.
It’s important to connect prediction with action. It’s useful to know that a shipment may be delayed, but the greater opportunity lies in using that insight to determine what to do next.
Nelson Teixeira, FedEx
To manage volatility better, the business is now using Demand Driven Material Requirements Planning (DDMRP). Where traditional Material Requirements Planning bases decisions on forecasts, DDMRP uses demand signals to decide what stock to hold, where to hold it, and when to replenish it. “Instead of asking, what do we think customers will need, we’re asking what are consumers drinking now, and how should we respond down the line?” Product flows are protected with strategic buffers that absorb normal fluctuations in demand and supply. These buffers are at different points along the supply chain and, when one hits a trigger point, it sends an alert that the buffer needs replenishing, he says. Machine learning can analyse the buffers and give recommendations around where they are placed across the supply chain and if they’re sized correctly, Anderson says. But there is still a staff member who can agree or disagree with these recommendations.
Anderson and his team have automated DDMRP using different technologies, such as demand-driven planning tool Intuiflow. CCBA also relies on Microsoft Dynamics and Power BI for ERP, dashboards and reports.
Other firms are also accelerating automation and investing in AI to streamline operations and mitigate disruption. At FedEx, improving efficiency across its network means closely monitoring operational signals. Every package moving through the FedEx network generates signals – from scan events and route progress to potential delays. By connecting these signals, it can better predict when a package will be delivered and manage expectations if the situation changes, says Nelson Teixeira, MD of operations for Sub-Saharan Africa. “Data and intelligence sit at the heart of our technology transformation. We are moving toward a more agile, cloud-first technology platform that brings together data, AI and modern engineering, and we intend to consolidate 100% of our data by the end of 2027.”
He says the goal is to move away from reactive logistics toward predictive logistics. For example, FedEx’s Global Delivery Prediction Platform is an AI-powered system that offers customers Predictive Delivery Estimates early in the buying journey. The platform uses historical delivery patterns, along with factors such as service type and public holidays, to provide customers with an expected delivery date at checkout, rather than waiting for tracking information after the item ships. This estimate is updated as the shipment moves through the network. “It’s critical to connect prediction with action. There’s value in knowing that a shipment may be delayed, but the greater opportunity lies in using that insight to determine what to do next.”
This accuracy is essential when working at the scale of a business like FedEx. “We move approximately 17mn packages a day and generate around 2 petabytes of data across 100bn transactions. Not all data needs to be processed at the same speed, which is why we focus our near-real-time capabilities where it matters most,” he says. “If we can improve an outcome, like more accurate delivery prediction or preventative equipment maintenance, that’s where this data really adds value.”
Even small adjustments can have a big impact. Mahendra Beharie, VP of IT for SSA at DHL Express, says every flight only has a finite amount of s pace and weight capacity, and DHL has to use every centimetre of space. Air cargo is packed into Unit Load Devices (ULDs), which are designed to fit a cargo hold.
“Because a box is generally square or rectangular, and an ULD usually has a rounded top, we need to be very clever about how we pack them to maximise space,” he says. “We use augmented reality to train our operations teams so that they learn how to best fit different types of boxes into a physical space.” To improve data capture, DHL uses OCR technology to convert commercial invoices into machine-readable, editable, and searchable digital text, which would previously have been captured manually. The business also uses picking robots in some of its high-tech warehouses that follow workers around the facility and carry heavy loads. “The goal is to use the technology available to us to work smarter and make better decisions,” Beharie says.
As tools and systems become faster and more sophisticated, many believe that more is better, but Anderson disagrees. “Yes, technology can help, but if you don’t have the right information at the right time, if you’re not spotting risks early and then making quick decisions that keep products available when customers need them, chances are high that you will lose a sale.”
* Article first published on brainstorm.itweb.co.za

