The paradox of route optimization: when theory collides with real-world operations
- May 14, 2025
- 5 mins
Despite its undeniable potential to reduce costs and CO₂ emissions, route optimization remains underused in the transportation industry. This paradox can be explained by the many practical challenges encountered during the implementation of such tools. From data quality issues to real-time disruptions, several factors contribute to the gap between expectations and the actual use of optimization technologies.
This article explores these challenges and highlights the adjustments needed to better align optimization tools with operational realities.
Route optimization: a major paradox?
Route optimization is all about finding the best way to visit a set of points while respecting a variety of constraints. It plays a crucial role in logistics, as a large share of transportation costs comes from the last mile—up to 53% of total costs in parcel delivery. This final leg also has a significant impact on CO₂ emissions, making it essential to optimize every aspect of logistics, both for economic and environmental reasons.
Route optimization solutions have been developed since the 1960s and 70s, offering numerous potential benefits, including:
Lower operational costs
Improved service quality
Semi-automated or fully automated planning
Reduced carbon footprint
What makes route optimization particularly fascinating for mathematicians is its inherent complexity. For instance, evaluating all possible combinations for a 20-stop route would take about 2,000 years—even with a powerful computer. With 60 stops, the number of possible solutions exceeds the number of atoms in the observable universe.
This complexity explains why well-tuned optimization algorithms can calculate better and faster routes than even the most experienced human planners. Given the sheer number of possible combinations, it’s impossible for a human to manually test them all. Algorithmic optimization consistently delivers significant improvements—often in double-digit percentages—whether in terms of distance traveled, productivity, or time window compliance.
However, despite these clear advantages and the maturity of this technology, its adoption in the transportation industry remains surprisingly limited.
This paradox raises important questions about the real-world challenges to implementing route optimization, despite its considerable potential to cut costs and reduce environmental impact.
Why is there such a gap between the need for route optimization and its actual use?
Many factors contribute to the paradox of route optimization, but they all highlight one central issue: data quality and data management.
Poor data quality
Data quality is a critical factor in designing delivery routes that truly reflect field reality. Unfortunately, poor data quality is a recurring issue that can seriously undermine the effectiveness of route optimization. Here are a few concrete examples:
Incorrect address entries: A small error in a street number can have big consequences. In Paris, for instance, a mistake on a long street like Rue de Vaugirard can result in a 300–400 meter discrepancy. With one-way streets, this can lead to major detours.
Faulty geocoding: When geographic coordinates are incorrect, it can lead to significant routing errors during itinerary planning.
Wrong weights and volumes: A simple data entry mistake—like forgetting a zero (10 kg instead of 100 kg)—can completely distort vehicle capacity planning.
Incorrect time windows: Errors in time settings can heavily impact scheduling, especially with traffic patterns and time-slot constraints. This applies to both drivers’ start times and delivery site opening hours.
In practice, data errors have a significant impact on route optimization. The generated routes can turn out to be unfeasible or illogical—for example, including unnecessary detours.
Unlike an experienced dispatcher who can instinctively spot certain anomalies, an optimization algorithm treats all received data as accurate. It cannot distinguish between correct and incorrect information. This limitation can lead to flawed outcomes when data quality is lacking.
Constraints only known by field operators
Route optimization algorithms often run into constraints that are known only to operational staff and are not documented in information systems. These field-based constraints can have a major impact on planning. Here are a few examples:
Last-minute changes: A driver may need to depart from home instead of the depot, or may be delayed due to a personal issue.
Interpersonal dynamics: Conflicts can arise between drivers and customers, resulting in situations where a customer refuses to be served by a specific driver—or vice versa.
Although crucial for effective planning, this information is rarely captured in the databases used by optimization algorithms. It falls into the category of tacit knowledge held by operational teams, built through daily interactions with drivers and customers.
Numerous real-time disruptions
Route planning is often challenged by unforeseen events that occur in real time. These disruptions can significantly affect the initial organization and impact operational efficiency. Here are some common examples:
Customer absences: It’s not uncommon for a customer to be unavailable during the scheduled time, requiring the delivery or service to be rescheduled.
Exceptional events: Unpredictable situations—like a presidential visit resulting in the closure of a major road—can completely derail pre-planned routes.
Last-minute changes: Customers may change their requests, cancel or postpone deliveries, or ask for new time slots—sometimes with very short notice.
These events have real consequences in the field. They disrupt planned routes, forcing drivers to adjust their itineraries—often without the support of optimization tools. This can lead to service degradation, with delays, cancellations, or rescheduling that negatively impact customer satisfaction.
Moreover, such unpredictability increases stress on teams, especially when routes are tightly optimized—since even a small disruption can place heavy pressure on operators tasked with managing these exceptions.
Ultimately, these constant changes call into question the relevance of pre-established plans. A plan designed the night before can quickly become outdated within a few hours, raising doubts about the effectiveness of long-term planning.
Highly specific business constraints
Route optimization frequently runs up against highly specific operational constraints, which can vary significantly across industries—and even between companies within the same sector. These specificities often make the use of off-the-shelf optimization tools particularly challenging. Here are a few examples:
Site capacity constraints: In waste collection or cross-docking operations, there may be limits on how many vehicles can be present on-site at the same time.
Loading/unloading constraints: LIFO (Last In, First Out) rules are often crucial in pallet transport, where only the most recently loaded items can be unloaded first.
Revenue balancing: When working with subcontractors, it’s important to distribute routes fairly to ensure each subcontractor’s economic sustainability.
Special access requirements: Some sensitive locations—like nuclear power plants or prisons—require specific driver clearances, which are rarely integrated into standard systems.
These constraints vary widely across sectors: waste management, temperature-controlled transport, parcel delivery, long-haul freight, etc. Each field comes with its own set of operational realities that must be addressed for effective optimization.
To account for such constraints, it is often necessary to build specific reference data—such as linking driver qualifications to delivery site requirements. Incorporating these types of constraints demands a more advanced and detailed modeling of the optimization problem, often making standard tools unsuitable.
The failure to account for traffic conditions
Integrating traffic conditions into route optimization algorithms is a major challenge. When not addressed, it can significantly affect the effectiveness of planning. Here are the main issues related to this omission:
Hourly traffic fluctuations not considered: Routes are often calculated without considering traffic variations throughout the day.
Post-optimization global adjustment: Traffic is sometimes added after optimization via a global coefficient, which fails to reflect the complex and dynamic nature of real-world traffic conditions.
The failure to account for traffic conditions in route optimization leads to several impacts on the field. Firstly, the Estimated Time of Arrival (ETA) becomes much less accurate, resulting in delays and missed delivery windows. Additionally, the algorithms may choose unsuitable and counterintuitive routes, such as crossing the Seine river multiple times within the same route in Paris, even though bridges are often congestion points, or using roads that are regularly congested at certain times.
These decisions can lead to illogical routes for experienced operators familiar with real-world conditions. Service quality suffers, negatively impacting customer satisfaction. If the system consistently produces results that don’t align with actual traffic conditions, it risks losing credibility with operators, undermining long-term adoption and effectiveness.
To address these issues, it’s crucial to integrate precise, real-time traffic data into the optimization process. This involves:
Considering traffic during route calculation: Using historical data to anticipate recurring congestion points.
Regularly updating routes based on real-time traffic during execution.
Additionally, optimization must account for other specific constraints, such as traffic restrictions (bridge height limits, vehicle weight, hazardous material transport, etc.), which can vary depending on the types of vehicles being used—from pedestrians to heavy trucks.
Integrating these elements helps generate more realistic and effective routes, improving ETA accuracy, service quality, and the tool’s acceptance by operational teams. This approach requires detailed, up-to-date mapping data from specialized providers (e.g., HERE Technologies).
Ultimately: A tool underutilized in the field
The practical consequences of route optimization problems are numerous and significant. Firstly, the routes generated by optimization algorithms often turn out to be unworkable in real-world conditions. This mismatch between theory and operational reality presents a major obstacle to the adoption and effective use of optimization tools.
Faced with these challenges, planners are forced to engage in a time-consuming and manual correction process. They need to identify errors or inconsistencies in the route plan, manually correct the data, and then rerun the optimization. This cycle repeats until the result is satisfactory, with each iteration taking between 5 and 15 minutes. Pressed for time, operators often abandon the process after a few attempts, thus frustrating the original goal of efficiency and simplifying route planning.
This situation frequently leads to degraded use of the optimization tool. In many cases, transport operators equipped with these tools end up using them suboptimally. The planner creates the routes entirely by hand, just as they did before the tool was introduced, reducing the optimizer to a mere calculator for kilometers and durations. This approach negates the expected benefits of optimization, both in terms of planning time savings and overall performance improvement.
It is also common that, to avoid these constant adjustments, operators tend to over-constrain the algorithm, forcing it to replicate existing route patterns. This practice severely limits the algorithm’s ability to discover new optimal solutions. By excessively constraining the algorithm, the potential benefits of optimization are lost, and the tool merely reproduces existing patterns without delivering any improvement.
In the worst cases, the failure to implement these tools can have serious consequences. It generates significant stress for teams, increases turnover among operators, and can lead to the complete failure of optimization projects. Management may then become entirely closed off to the idea of route optimization, considering that it simply doesn’t work.
These difficulties largely explain the low adoption rate of route optimization tools in the industry, despite their theoretical promises. This situation highlights the need for optimization solutions better suited to real-world conditions, capable of integrating all the specific constraints and adapting flexibly to real-time changes. A more holistic approach, taking into account not only the technical aspects but also human and organizational factors, is essential to overcome these obstacles and fully realize the potential of route optimization.
To address these challenges, Kardinal has developed a next-generation route optimization solution, functioning continuously for better user adoption. Discover its innovative features and see how it works with concrete examples in our upcoming article, ‘What solutions can improve the use of route optimization tools?