Logistics Demand Forecasting: The Benefits of AI & How to Implement It
For many logistics companies, the road to digital transformation and AI implementation is not an easy one. In an industry that has largely been run by pen, paper, and phone for decades, the transition to using modern software and tools can seem challenging and even overwhelming. What many of these companies don’t realize, however, is that they are creating an even bigger challenge for themselves by not implementing some of this cutting-edge technology into their operations.
Companies that don’t use logistics demand forecasting find that it makes the operational planning of assets very difficult. The multi-faceted problem requires businesses to consider how many assets they need, whether or not those assets are positioned correctly at any given moment in time, and how to best plan the technical breaks.
This is a very complicated problem to solve, as it requires a large volume of interdependent information. Luckily, logistics companies already generate a tremendous amount of data internally and have access to even more data from public sources. Nevertheless, the challenge remains that only a few tools currently exist which allow companies to synthesize all of this information and enable data-driven decision-making in conjunction with the experience and instincts of their managers.
But with the help of modern predictive optimization tools, logistics companies can shift to an anticipatory strategy based on accurate demand forecasting, and thereby achieve far greater operational efficiency. Let’s take a look at what exactly logistics demand forecasting does, how it works, and its many benefits for logistics companies.
What is Logistics Demand Forecasting?
Logistics demand forecasting is a way for companies to accurately anticipate the demand for products and shipments throughout the supply chain, even under uncontrollable conditions or circumstances. To accomplish this, it helps for logistics companies to implement a forecasting model to predict capacity demand, relying on a combination of their own historical data and multiple external variables, such as the one developed by Transmetrics. The model should allow for the manual adjustment of forecasts in order to account for new customers or other changes in business and increase accuracy.
By creating and implementing their own personalized demand forecasting models, companies can more easily achieve an accurate forecast in a number of different ways. These models can help companies better understand exactly how much safety stock they need, or the level of extra capacity the company needs to facilitate in order to meet the unexpected demand. Additionally, using logistics demand forecasting models can help decrease the number of kilometers spent repositioning assets, increase cargo vehicle capacity utilization; and increase asset utilization for asset owners including shipping lines, trucking, and intermodal companies.
In general, there are two types of logistics demand forecasting that a company can model: medium and long-range forecasts, and short-range demand forecasts. Medium to long-range forecasts are strategic. In this scenario, companies typically use the data for budgeting purposes and planning purchases of new assets such as trucks/ships, warehouses, distribution centers, and building new hub facilities. These demand forecasts can range from anywhere between half a year to three years.
But the most useful type of demand forecast for logistics companies typically comes from the short-range type. This highly influences the operational planning and helps to improve the bottom line for companies with low-profit margins. Short-range forecasts generally predict a few days or a few weeks ahead. For example, they are up to 14 days for ground freight transportation and one to 12 weeks for ocean shipping. Let’s take a deeper look into a few of the ways that short-range forecasts can create a wide array of new benefits in a logistics company’s supply chain.
Subscribe to the Transmetrics newsletter ✉
Logistics, AI, and Tech analysis. Monthly in your inbox.
Optimizing Operations to Reduce Costs
Logistics companies that don’t use modern technology can end up struggling with lower profit margins due to unnecessary costs. Half-empty trucks or containers and inefficient operations and maintenance are just some of the factors that can end up costing logistics companies more than they realize. However, by properly implementing logistics demand forecasting models and taking action based on these advanced technologies, companies can optimize the supply chain and help to cut costs across the board. How? These solutions can offer benefits like the reduction of fleet sizes, leasing costs, maintenance costs, storage costs, parking, and driver costs.
Let’s take a look at how AI and data have made an impact on the courier company, Speedy, part of DPDgroup. The company controls one-third of the courier, express, and parcel market in Bulgaria, with 16 million parcels handled every year. However, the e-commerce boom found the logistics operator needing to invest heavily in modernization practices. The introduction of Transmetrics software, along with other initiatives of the management team, allowed Speedy to quickly identify and cancel unnecessary trips and line hauls under the new system. This led to a 25% hub-to-hub cost reduction, as well as a 14% increase in fleet utilization.
Overall, the technology helped Speedy to cut costs by an estimated seven to nine percent. Cost-cutting like this is a win-win: their operations will be more efficient for less expenditure. For logistics companies, this means that profit margins can rise when those unnecessary costs are eliminated and demand forecasting is accurate.
Increase Employee Efficiency
Using data analytics can simply make life easier for employees, helping them to perform more accurately and timely than before. If logistics planners have all insights and data, they will have more time to spend on operations rather than on trying to calculate and predict where they need to locate their assets and check how full they are. For example, AI algorithms can calculate granular demand forecasts on each origin-destination. Then, after capturing final decisions from the planners, algorithms can compare the outcomes between human and AI suggestions, and use this input in order to train the AI models further. With this approach, logistics providers can keep their experienced employees in the planning process, leave them in control and use their knowledge to improve the AI. A higher reliance on tech to make data-driven decisions can make the employment much more efficient for both parties.
There’s no doubt that employment accounts for a big chunk of company expenses. For logistics planners, using data-driven technologies such as demand forecasting is a sure-fire way to increase employee efficiency and create better use of time, resources, and salaries. Without data, companies may not know whether to let a driver take his weekly break immediately or wait until he arrives at his next destination. However, using data can provide a definitive answer to that question while reducing costs.
Optimal Fleet Repositioning
This idea is simple, but not always so easy to implement: ensure all trucks, containers, and spaces are utilized to capacity. Take the example of TIP Trailer Services: The company manages a transport fleet of more than 70,000 units over 70 locations in Europe. With so many locations and so many units, the company wanted to achieve a more efficient fleet repositioning schedule. This is where they solicited Transmetrics to assist in predictive analytics. Transmetrics applied their software solution to cleanse and enrich the company’s historical data, after which they implemented an AI-driven demand forecasting model which was tailored specifically for the TIP Trailer Services business case. The result? TIP Trailer Services could predict asset demand two weeks ahead with 98 percent accuracy, and six weeks ahead at 95 percent accuracy. This means they were able to implement one-way rentals and offer their customers a significantly more flexible service while anticipating a potential 11 percent increase in revenue.
Selling Extra Logistics Assets
Making money on all available space is what logistics companies should strive to achieve. It appears that not all companies are capable of this feat, however. For example, The World Economic Forum reports that half of the trucks travel empty on their return journeys following delivery in the European Union, a clear waste of assets.
One major part of underutilization is the fact that logistics operators need to have a safety stock of assets everywhere. As a result of optimizing their operations with proper demand modeling, companies could sell off extra assets that they don’t need and in turn reduce inventory costs. Or, companies with accurate forecasts can use sub-contractors at a cheaper rate once they know weeks in advance that the demand is predicted to be higher than usual. In this case, the company does not need to purchase and care for the extra assets as safety stock, but rather identify peak periods and lease the required assets for the required time.
Next to consider is dynamic pricing. This is all about getting the best return on investment based on the current supply, demand, and market status. This practice can ensure logistics companies sell their goods and services for the right price at the right time. Logistics demand forecasting, combined with a solid understanding of capacity and inventory, allows companies to better scope prices and where they should be set. This is a vital component in the competitive world of logistics where every company is attempting to outprice their competitor. Those who know what they can realistically offer will have an important headstart – yet another reason to create a demand forecasting model.
A Final Word
Now that we know more about the topic, it’s crucial to take note that short-range forecasting will not help a logistics company on its own. Even if a company has this accurate demand forecast for hundreds or even thousands of its origins and destinations, it’s unlikely that their planning teams will be able to fully benefit from all this extensive information — at least, not without using additional software tools to suggest some real actions and support decision-making based on this information. In fact, it’s imperative that companies use demand forecasting in combination with Augmented Intelligence tools to optimize their supply chain and make their business more efficient than ever before.
Combining augmented intelligence with demand forecasting could be a new recipe for success for most logistics companies: Consider that in mere months from implementing a demand forecasting model and AI tools, logistics companies can usually expect about a 15% reduction in the fleet of assets, 10% decrease in the cost of logistics of empty assets, and 15% savings in the asset maintenance and repair costs.
Most people are now accustomed to hearing the phrase “artificial intelligence (AI)” — but there’s a different type of AI that’s growing just as popular. This newer form is called “augmented intelligence,” and it’s ideally used to create a “human-in-the-loop AI” approach. This approach attempts to combine the best of both worlds — using artificial intelligence to gain new insights or recommendations for actions to take, but including a human in the process to actually decide if that action should be taken or not. The human-in-the-loop approach is possibly the most effective form of augmented intelligence, but can also be the most difficult to achieve. But for logistics professionals, this form of augmented intelligence is the one to strive for if the idea of turning over some parts of the logistics planning into artificial intelligence still seems a bit scary. The concept is becoming so important that we have written an extensive article on the topic so every logistics professional can understand it in a better way.
Clearly, logistics demand forecasting brings together both short-term and long-term benefits for logistics companies, and deciding to use specialized tools to create forecasting models is arguably one of the best choices that a logistics management team can make in today’s competitive supply chain environment.