Choosing a route optimization solution is often seen as a purely technical exercise, based on objective and measurable criteria. Yet in practice, this process turns out to be far more complex and nuanced than theory suggests.
Between standardized benchmarks and detailed RFPs, companies often multiply the “classic” approaches to compare tools available on the market. These methods may seem rational, even indispensable, but they often miss the point: route optimization is not just an algorithm to evaluate, it’s a human, operational, and evolving project.
In this article, we explore why these standard approaches can be misleading and propose a pragmatic framework for making an informed choice, centered on real-world operations, collaboration, and long-term adaptability.
What does a typical route optimization procurement process look like?
When a company seeks to implement a route optimization tool, it often adopts one of two traditional approaches used for business software procurement. While these methods appear rigorous and objective on paper, they are often poorly suited to operational realities.
Case 1: The “benchmark” approach
In this first scenario, the buyer provides vendors with a dataset they consider representative of their real-world activity—typical routes, covered areas, volumes, logistical constraints, and so on.
Each vendor is then invited to “optimize” these data using their own algorithms, and the results are compared based on quantitative indicators such as:
- route calculation speed,
- total distance traveled,
- vehicle fill rate,
- or estimated productivity.
This approach aims to create an objective technical comparison, with the goal of selecting the solution that “performs best” under a given framework.
Case 2: The RFP with a functional requirements list
In this second approach, the company prepares a detailed Request for Proposal (RFP) describing an exhaustive list of functional requirements, business constraints, and technical specifications.
This list might include, for example:
- types of vehicles to manage (trucks, bikes, heavy goods vehicles, refrigerated vans, etc.),
- delivery or collection time windows,
- customer-specific constraints (reception slots, priority levels, accessibility),
- business rules specific to the organization (break times, rest rules, driver qualifications),
- and expected performance indicators (on-time rate, cost reduction, carbon footprint, etc.).
The goal is to obtain precise responses from vendors demonstrating their functional compliance with the specifications.
Often, this RFP is combined with the benchmark approach to cross-check a qualitative assessment (features and compliance) with a quantitative one (algorithmic performance).
On paper, these two methods have everything to reassure. They help structure the evaluation, avoid decisions based solely on price, and align stakeholders around a “scientific” selection process.
Yet in practice, these approaches pose several problems, specific to route optimization. And it is often at this stage that future project difficulties begin to emerge…
Why these approaches are not relevant
Both the benchmark and RFP approaches, while traditional, have major limitations when it comes to route optimization.
Limitations of the benchmark approach
The main issue with benchmarks is that the provided dataset is often too simplistic. For practical reasons, data is standardized, secondary details are removed, and the real complexity of operations is smoothed out. This results in an almost academic problem, stripped of the subtleties of real-world operations.
On such datasets, the performance of different solvers is often virtually identical. Any remaining differences may come from vendor-specific tricks (unrealistic speed assumptions, simplified constraints, etc.), but can also result from the fact that vendors use different providers or sources for their distance matrices. In such cases, results become completely non-comparable, regardless of the actual quality of the algorithm.
Moreover, strong performance on a simple dataset does not guarantee quality on a complex problem. A poorly modeled or ignored business constraint can quickly cause results to collapse.
Conversely, some vendors invest heavily to create a highly realistic dataset, closely matching operational reality. In this case, it becomes impossible to objectively compare the performance of solutions: each vendor models constraints and objectives differently. Often, the vendor who accounts for the most complex constraints ends up further limiting the problem, which may appear as lower optimization performance, even though the solution is more faithful to operational reality.
Limitations of RFPs
Typical RFPs often resemble a laundry list of constraints and requirements, with many generic points and little focus on the real operational challenges.
Precisely modeling these business constraints is a discipline in itself, the domain of optimization experts and mathematicians. It is not the job of operations staff, IT, or procurement teams.
Thus, an RFP can only provide an approximate framework, often incomplete, which misses critical points. All vendors tend to answer “yes” to most requirements because they are too vague or simple.
Meanwhile, the reality of constraints is continuously evolving, between company strategy and day-to-day operations; this makes a fixed list even less relevant.
Comparison table: benchmark vs. RFP approaches
Aspect | Benchmark approach | RFP approach | Description | Providing a representative dataset for each vendor to optimize using their own algorithms. | Publishing a detailed requirements document listing functional needs, business constraints, and technical specifications. |
|---|---|---|
Objective | Compare algorithmic performance on the same dataset (speed, distance, vehicle fill rate, productivity). | Obtain precise responses from vendors on their ability to meet the listed requirements and constraints. |
Advantage | Objective, quantitative, rational technical comparison. | Complete and detailed functional evaluation of the solution’s capabilities. |
Main limitations | – Dataset often too simplified, not representative of reality. – Results often very close or even biased. – Difficult to compare if datasets are highly realistic and constraints are modeled differently. | – Requirements list often too long and generic, with little focus on real operational challenges. – Complex business modeling rarely mastered by the buyer. – Inflexible in the face of continuously evolving real-world constraints. |
In summary, these “classic” approaches neither predict the real success of the project nor accurately evaluate a vendor’s ability to adapt to the client’s specific and evolving challenges.
This is why it is necessary to go beyond benchmarks and RFPs: understand the business subtleties, involve operational teams, and above all, adopt an iterative dialogue with vendors to co-create the right solution.
Our recommendation for choosing a route optimization software
Selecting a route optimization vendor should start simply, while maintaining high expectations over the long term. Today, the technical quality of solvers is generally quite homogeneous. What really makes the difference is the ability to accurately model the problem, integrate the solution into operations, and, above all, support the client over time.
We recommend a pragmatic three-step approach:
Meet several vendors to identify the one with whom human interactions work well and whose operational teams collaborate smoothly. This is fundamental, as success will depend on daily collaboration and iterative adjustments.
Start with a pilot on a limited scope, without long-term commitment. This phase should allow you to “crack” the problem and validate the value delivered before committing fully. If your project team is large (5–10 people), you may consider piloting two solutions simultaneously, but this is rarely feasible otherwise.
Treat route optimization as a strategic and iterative project. The vendor should be able to guide you through multiple improvement cycles, with patience and responsiveness. It is also essential that they have prior experience in contexts similar to your business, to avoid misunderstandings that can waste valuable time.
Continuous route optimization solution
Conclusion
Choosing a route optimization solution is not simply a technical exercise or a paper-based comparison. Classic approaches, benchmarks and RFPs, while useful to structure the procurement process, quickly show their limitations when faced with the complexity and diversity of real-world needs.
Route optimization is above all a human and collaborative project, requiring a deep understanding of business constraints, data quality, and the operational environment. Only by prioritizing dialogue, concrete testing phases, and an iterative approach can a company select a solution that is truly adapted and sustainable.
Ultimately, success depends less on the raw performance of an algorithm than on the ability to build a strong partnership with a vendor capable of supporting the project’s evolution over time. Simplifying the initial choice, focusing on collaboration and flexibility, this is the key to turning the promise of optimization into tangible and lasting gains.
💡Read more: What you should never overlook in route optimization
💡Want to learn more about our route optimization solution?
Contact us for a demo and discover how Kardinal can support your organizational transformation.