top of page
AI model for credit risk reduces workload and speeds up customer response times at Boels Rental
  • Accepting new customers for rental on credit was a bottleneck, which created an accumulating workload for employees

  • The AI model supports an immediate decision in more than 60% of the rental requests, leading to substantially shortened review times, faster response to the customer and a reduced workload for the team

  • Seven weeks for project design, development, testing and implementation

At a glance:

Boels is one of the largest equipment rental companies in Europe. This includes both equipment rentals and specialist rentals. On a daily basis, Boels receives approximately 150-200 requests from new customers who wish to rent on credit. Each new request had to be reviewed manually by an employee.

The credit review process proved to be a bottleneck in accepting new customers for rental on credit. Customer requests accumulated, resulting in a higher workload for employees and longer response times towards the customer than desired. Boels took steps to automate the review process as much as possible and was looking for ways to automatically filter out (part) of the cases that can be immediately accepted or rejected. The goal was to reduce labor efforts, improve decision quality and up the speed of servicing customers.


“The original expectations have been exceeded, as in more than 60% of the cases the model can directly support a decision. Our employees are very happy with this, as customer requests are handled substantially faster than before. Also, it is beneficial for our team as we can focus on other tasks with the same capacity.”

Bjorn Curvers | Manager Credit Risk at Boels Rental


Creating a model with the ethical use of data

Besides the frequent cooperation with the finance team, other relevant stakeholders were actively involved resulting in a multidisciplinary project team with members from IT, analytics and OutSystems developers. First, we cleaned, structured and explored the data to create predictive variables to train the AI model. For this, we also used basic text mining to extract relevant but hidden information that could be predictive of a credit risk decision. During this process, we took care to avoid using any customer characteristics that may lead to undesired bias by the model, which is our standard procedure to ensure that only information directly relevant to the decision-making process is included. Several AI models were trained, tested and validated. Thereby, we stayed close to the existing workflow of the finance team to match the model to the process steps where information would be available, which is crucial for successful implementation. Validation of test data indicated we would exceed the expectations of the original business case, which led to the implementation of the model in the real-world process.


“Thanks to the frequent and efficient collaboration with relevant business owners and the analytics team at Boels, everyone on the team looks back on a successful project.”

Rinke Klein Entink | Director Data & Analytics at IG&H


Validation and steady progress to win the AI race

After technical implementation, the rollout of the model occurred in stages. First, a user would still validate 50% of the model results manually; to build trust in the model and detect unforeseen situations. Gradually, this was brought down to 5% of randomly selected cases for manual review, to make sure the model stays up to date. End-user acceptance of the model was a top priority. Therefore, we trained the users and had them experience the workings of the AI model so they would understand and trust the outcomes.

Employees embracing AI

The slight initial uncertainty about developing an AI model quickly faded as employees saw how the obvious (credit) requests are simplified and require much less manual effort. The AI model supports the process in more than 60% of customer requests. Besides volume, it also increased speed: As most customers receive an immediate, automated decision, this resulted in a substantial reduction of the response time towards Boels’ customers.

Through guidance during the process coupled with attention to implementation and change aspects for the end user, the project was successfully completed within a short timeframe that met the planned deadline. Boels is now successfully using AI in supporting decision-making processes at scale, without compromising quality and simultaneously improving its service to the customer.

Read more retail client cases

Ready to transform your business?

Stay ahead of your competitors by transforming your business. We have more than 30 years experience in helping clients innovate, deliver superior experiences and execute on their strategies. Curious how we can help?