Pre-operations

1. Capacity and resource planning
   People are the most important asset in logistics but are becoming scarce. Therefore, having high accuracy in estimating the right capacity and the right people with the right skills for the coming period is crucial. Using new mathematical techniques, you can train your forecasting model for better accuracy and compute the resources you need, such as the number of shifts, required capabilities, and timing. Over the longer term, machine learning models can also support recruitment by determining the ideal mix of contracts and capabilities based on seasonal patterns and fluctuations over the week, including own resources and those from temp agencies. The technology can also be applied at the operational level to define healthy rosters, thereby improving employee satisfaction.
2. Steer your customers
   Use dynamic pricing and revenue management techniques to guide your consumers to the right time slot to maximize clustering and efficiency. This is already common sense in online grocery deliveries when choosing a certain time slot but can be applied more broadly. For example, defining delivery days and frequencies in food service distribution or implementing it in the dynamic world of last-mile delivery. By using simulations and machine learning models to learn from human behavior, you can analyze the consequences before implementing changes in practice and apply A/B testing.

Post-operations

1. Cost-to-Serve
   Machine learning allows you to assign costs to all activities more accurately than in the past with Activity-Based Costing. In logistics operations, such as transportation, you can define features (e.g., volume, distance, proximity) and then use details (e.g., which orders are combined in one route or loading) to train a model that assigns cost to each element. By comparing these costs with the associated revenues, you gain a clear view of profit- and loss-making activities. This insight is helpful in determining where to focus your energy, whether to extend or end contracts, where to apply promotional activities, and the impact of adding new customers or activities. Machine learning opens up a broad range of new opportunities.
2. Improving master data, including stop and driving times
   When running an operational and execution system in logistics, you can collect a large amount of data, such as stop and driving times. By using Machine Learning, you can understand the factors behind a certain stop time, such as specific address aspects, volume, driver, and others. Similarly, you can improve the expected driving time per road segment for each moment of the day, taking traffic congestion and incidents into account. This not only improves the reliability of the estimated time of arrival (ETA) but also leads to greater efficiency. If the driving time was too slow in the past, this knowledge directly contributes to improvement, while if it was too fast, the optimizer will look for alternative ways or moments to deliver these addresses. Machine learning can also be applied to other aspects, such as learning how pallets or containers are stacked in practice or understanding the expected cost of outsourcing external logistic platforms that use dynamic pricing.

Conclusion

These are just four practical examples of how Machine Learning can improve your logistics operations, but I believe these opportunities are also valid in other industries. It is great to see the value in your data, learn from it, and improve your service and efficiency smoothly. This creates a learning loop of continuous improvement and fosters a culture of excellence within the company. Furthermore, it can be used to enhance your sales and support growth strategies. A new era of opportunities has started, harnessing the power of machine learning and mathematics.