Scoring

Summary

Scoring is a way to measure how likely a client is to repay their debt. It’s represented as a number, making it easy to assess someone’s financial credibility at a glance.

Our system calculates this score using Machine Learning. Trained on over 1.5 million features with a mix of statistical models and algorithms, it provides accurate and fast results. This enables you to quickly and confidently decide whether to approve or reject a loan application.

How Scoring is determined

Scoring is built around a Machine Learning model that predicts the probability of a client repaying their obligations. To train the model, we use labeled datasets from our partners that indicate whether or not an end-user’s repayment occurred. This process is essential - it helps the system learn patterns and relationships in the data that correlate with repayment behavior. The model learns to recognize what a “good” client looks like in terms of repayment potential.

Once the end-user’s data is on the Kontomatik servers (e.g. from AIS, PDF documents, or your own dataset) our system can generate:

• Score: Probability of repayment, represented on a scale from 0 to 1 (higher is better). A score of 0.8 means that there is an 80% likelihood that the user will pay off their debt.
• Score Percentile: This shows how the end-user compares to other users applying for a loan based on our training datasets. For example, a score in the 0.90 percentile means the client is more likely to repay than 90% of all users that were included in the model training process.
• Tier: This simplifies decision-making by grouping clients into easy-to-understand categories (letter A-F). For example, Tier A represents clients with the highest likelihood of repayment, while Tier F includes higher-risk individuals.

Supported services

The scoring endpoint requires an ownerExternalId as input, meaning it can calculate scores using all data aggregated for a single end-user. Supported data sources include:

• Account Information Service
• PDF Statements Parsing
• Owner Upload

Integration

Integrating our scoring system is very straightforward:

1. Import end-user data
2. Call the /owner-scores.xml endpoint. Pass the ownerExternalId assigned to the end-user.
3. Save the data from our API

The API will return:

• Repayment probability – a numerical value representing the likelihood of timely repayment.
• Percentile – how the client compares to others in the dataset.
• Tier – a classification (letter A-F) to simplify decision-making.

For more technical details, refer to the Scoring section in our documentation. There, you’ll find everything you need to implement the solution, from endpoint specification to example API call.

Custom Scoring

Kontomatik can create a custom Scoring model, tailored to your specific criteria. This model will offer a unique structure of responses and values, which will be explained to you during the preparation process, ready for your use.

Additionally, Kontomatik provides Owner Features - a comprehensive set of Machine Learning-based metrics that are used in our scoring models and describe the profile of the account owner. In case you don’t want to rely solely on our final score, you can integrate selected features into your own decision-making process.

To access either of these options, please contact our Data Science team at data@kontomatik.com for further details.