[REPORT]
Data in route optimization: state of play, issues and best practices
In the previous article, we identified key data required for truly relevant route optimization. Among them, the task orders, the available resources and the specific constraints of the business are essential information collected by the client. Thanks to the digitalization of logistics and the transport industry, it is now easier to collect and store this data. However, the challenge is not only to collect as much data as possible, but also to ensure that it is of high quality.
This issue has been coined in the expression “Garbage In, Garbage Out” (GIGO), which describes the fact that the input of incorrect data will lead to meaningless results. GIGO applied to Artificial Intelligence and Machine Learning has proven on several occasions that human biases could impair algorithms that were meant to be objective during automated recruitments, reproducing existing discriminations.
Low quality or incomplete data
This concept is also essential in route optimization: With poor data, the calculated routes will be unusable. Thus, a mistyped address will disrupt the relevance of the route: the vehicle might park at the wrong location, the driver will take more time to find the right address and this will lead to delays. Delivery drivers often have to deal with these problems on their own, even though they could have anticipated them.
In logistics, it is very common to have missing or incorrect data: an inaccurate address, interchanged fields (e.g. phone number in the address field), wrong package weight or size, etc. This quality issue remains widespread today since operators frequently enter data by hand, increasing the risk of error. Carriers are also dependent on the data collected by their shipper customers, which is often missing or incomplete.
Data on task orders and available resources are usually collected, but not data detailing specific business constraints. These are known by operators, available in their heads only. For example: a customer has a preference for a given operator, a driver is late this morning, another is faster in an area he is more familiar with than others, etc. If the route optimization software is unaware of this essential information, it won’t be able to plan the best routes possible.
In the case of routes with technicians or experts, skill data is often not up-to-date due to the high turnover in the industry. However, this data is essential, as technicians may have different skills that are not always relevant to each customer. Depending on the complexity of the task, the time spent by a technician on a customer’s premises can be very different. Without this information in the system, poor routes are created with inaccurate times and wrong technicians.
Data can cause prevent optimization projects
Today’s logistics players are aware of the importance of data and its impact on their business if it is of low quality. Digital maturity is gradually becoming the market standard and logistics companies are starting to look after their data.
The majority of Kardinal’s clients have started to improve the quality of their data, and some are in the process of completing the transition. In the past, we have had many projects terminated due to lack of or poor data quality.
Cédric Hervet, co-founder and Head of Science at Kardinal
Companies that have grown externally, i.e. by acquiring other companies, find it difficult to standardize and centralize their IT systems. The data, generally scattered in Excel files, is not always usable by the optimization software.
We encountered this situation during discussions with an insurance company. They needed to optimize the movements of their experts who were traveling to perform audits. However, it was almost impossible to automate the optimization because of missing data: no information on the task orders, on the skills needed to perform these audits, there were inconsistencies in the data…
In these cases, the route optimization project often helps the client understand its lack of digital maturity and realize that a lot of work is required to collect and sort the data.
We also had to postpone a project with a parcel delivery company that had tried to standardize data transfers via EDI (Electronic Data Interchange) so that carriers could share transport orders. However, the integration was flawed (address fields were incomplete, fields were interchanged, etc.) and would have required manual reprocessing of all of these EDIs every day, which was impossible.
Data quality is now a key issue, which implies a real work on digitizing logistics processes. Many companies in the industry do not enter all their data into their systems. Key data is still incomplete or not accessible to the optimization software, which then calculates unrealistic routes in the field. This leads to a significant loss of performance. So, what is the solution to this data issue? How can we improve the quality of the data and maintain it over time? Find out more in the following article >>