Introduction
Logistics costs play a critical role among operational cost items for companies, with an ever-growing importance as the scope of operations is becoming more complex. Logistics has conventionally been one of the main pillars of operations in manufacturing industry, yet, while value chains in almost all sectors tend to get more sophisticated, importance of logistics operations mounts for all companies.
Nevertheless, many companies are inclined to outsource logistics operations due to several important factors.
Amongst them, physical expansion of operations, logistics-specific know-how requirements, heavy-asset necessities, involvement of macroeconomic and political factors are prominent reasons for companies to outsource such a fundamental process. However, the uncertainty brought by these factors are at the center of the business model of 3rd Party Logistics (3PL) service providers. These companies tend to leverage this uncertainty and complexity of operations to maximize their earnings.
Given the intricate nature of operation and a multitude of input parameters, it is often difficult for companies to come up with an accurate estimation to form a basis for planning projections and budgeting. Majority of companies lack a data-driven decision support system to increase their control on the logistics costs, predict logistics costs based on historical data, and manage logistics tenders and negotiations more effectively to reduce logistics procurement costs.
The logistics service category has the highest amount of expenditure in the indirect purchasing categories of companies today and in the next five years. More accurate estimation of logistics-related expense budgets in the coming years according to the dynamic and changing conditions of this category is of critical importance in order to prevent the shrinkage of profit margins due to the uncontrolled increase in operational costs.
In order to fill this gap, the analytical approach developed by PwC targets these pain points with an AI-powered solution. On the contrary to conventional transportation management solutions, artificial neural networks as the basis of the AI-engine successfully captures relationships between potential cost drivers and pricing mechanism of 3PL service providers. Therefore, the AI-engine takes not only the client’s tender specs such as loading location, destination, volume and etc., but also external factors like fuel prices, infrastructural connectivity and economic activity into consideration.
Based on the client’s historical data for logistics tenders and aforementioned external factors, AI-engine predicts cost intervals for logistics tenders with unprecedented accuracies compared with conventional approaches. As the AI-engine is built up with the client’s data, it adjusts its design exclusively for different logistics procurement categories (e.g. international/domestic road transportation, container transportation, storage etc.) in the client’s business model. Yet, besides the client’s data, external data included in the model by highly credible sources such as UNCTAD, World Bank and OECD significantly increase the robustness of the model.
Business Case
Logistics cost estimations were carried out for a leading industrial manufacturing company using the cost estimation tool developed by PwC. As a part of this project, 5 different AI-engines were developed and delivered for 5 different modes of transport (domestic inner-city, domestic inter-city, international road transportation, maritime container and bulk transportation). All models were delivered supported by training documents and user manuals exclusively prepared for the company.
PwC’s supply chain consulting team prepared a longlist of client data. After dialogues with experts, external parameters representing the external factors in 3PL service providers’ cost structure were also listed up. After gathering a pool of potential parameters, an impact assessment to filter the ones having a significant impact on the costs was conducted.
Thereafter, an AI-engine with artificial neural networks was built after sessions of training and testing of the design. Later, the design was simulated with a renowned statistical method called Monte Carlo Simulation to identify real-life potential intervals for tender prices. Finally, the software design was completed on both Python and R.
How Can We Help as PwC?
PwC’s procurement solutions continue to set examples of best practice in the field.
Contact us to estimate your logistics costs with a systematic approach based on analytical foundations, achieve savings higher than your targets in your logistics tenders and to get more detailed information on this subject.
