In logistics, empty trucks drive somewhere between 15% and 38% of the time. In 2024, 21.6% of all EU truck kilometers were driven completely empty – and that doesn’t include partially empty ones. It’s been a problem for as long as freight has existed, and most companies treat it as an inevitable cost.
It isn’t inevitable. There are better ways to manage fleets. Before, it was a significant challenge, but now, we have artificial intelligence, which excels at matching available loads to available trucks across constantly shifting routes.
Why it’s still happening
The obvious question is why, if this costs the industry a fifth of its productivity, nobody has fixed it already.
A dispatcher managing hundreds of trucks across multiple countries is dealing with constantly changing load availability, delivery windows, driver hours, border crossings, and fuel costs. The human brain can’t take in that much information all at once. Finding the optimal route manually isn’t slow because people aren’t trying hard enough, it’s slow because the number of possible combinations is genuinely overwhelming.
Static route planning tools help, but they can’t adapt in real time. A load that becomes available in Warsaw at 3pm doesn’t fit neatly into a plan that was built at 8am. Dispatchers have no other choice than to let the trucks drive empty, and the focus shifts on to the next problem.
How we’ve fixed it before
A logistics company we worked with was dealing with this exact problem. They had a large volume of freight moving across European routes, with trucks regularly completing deliveries and returning empty.
The first challenge was acquiring data. We pulled together over 15,000 freight entries from Lithuanian and European-wide cargo platforms, and analyzed the load availability, distances, routes, pricing, journey times. The goal was to build a system that could look at where a truck would be after a delivery and find the best available load for the return.
The algorithm we used is called Ant Colony Optimization. The name sounds more exotic than the idea: it mimics how ants find the shortest path to food by laying down pheromone trails that other ants follow and reinforce. Applied to routing, it simulates thousands of possible paths simultaneously, and gets better as it runs. Unlike a static planning tool, it adjusts as conditions change, meaning it can consider new loads, updated locations, shifting capacity.
The system connected to the client’s logistics platform via API, so route recommendations updated automatically based on live data rather than a morning plan that was outdated by noon. A single backhaul load can nearly triple the profit on a given route. Across hundreds of routes, the numbers add up quickly.
What this means for your business
Empty trucks are not an unavoidable logistics problem. They’re a data problem: there are too many variables that are moving too fast for any dispatcher to track manually. The technology capable of solving it already exists, and not using it means significant losses for the company.
If your trucks are regularly completing one-way trips, it may be time to consider whether route optimization could help. The investment into the system is only a fraction of the money lost without it.