Our approach

So, how does AdviceRobo work?

We take your applicants’ and customers’ behavioural data, and the results of their psychographic testing, and paint a clearer, more defined picture of them. This can be in relation to their creditworthiness, their vulnerability, the likelihood that they’ll churn or practically anything else you want to predict when it comes to credit risk. Each of our products are based on machine learning technology to achieve this goal.

Data that gives you the complete picture

We combine the power of psychographic data, browser data, user interaction tracking data, transactional data and are continuously adding new data sources.
This creates a rich data set that our software robos can use to create accurate predictions.

User interaction tracking
Financial skills
Debt attitude
Social desirability
Interaction effects
Time between invite and start
Time spent PCS
Time spent per question
Browser type
Location country
Browser version
Location region and city
Location Zip code
Number of clicks
Mouse movements
Mouse hover over
Number of scroll movements
Mouse time
Out of focus
Clicks on answers
Mouse distance

No personal information is stored in our database, according to the GDPR rules.

Get the insight you need with our Deep Learning AI

The actual data
Traditional data
Our data
The real picture

In real life data is complex and non-linear.

In this example the customer is 25 years old and has a Masters degree. They’re a fashion industry employee who receives regular variable remuneration. They come from a wealthy family.

The traditional view

Using traditional regression statistics the data view follows a pre-determined range which is the case for traditional risk scoring models.

The range when mapped to real-life data often includes bad risk yet excludes good opportunities.

With a traditional risk scoring model, this customer would probably have been rejected.

The Machine Learning view

Using Machine Learning algorithms our robos can find accurate predictive patterns, providing a much more accurate view. This allows you to better identify risk and opportunity.

With machine learning and psychographic testing this customer would fall into the approved category.
The Robo takes into account more than just the basics – the deep data dive allows for a more comprehensive check meaning more approvals.

AI driven risk prediction models

Our metrics
Model development

The quality of the AdviceRobo models will be measured with:

  • Accuracy
  • Precision
  • Recall
  • F1

The AI driven risk prediction models are ready to use but can be tailored towards the specific lender. This depends on the quality and quantity of the available customer behavioural data.
The models are developed following a strict procedure and will make optimal use of the available customer data at the lender. Also, Psychographics could be added to the models and so create a much richer customer profiling.

All services are delivered globally via the AR platform. This platform is at the heart of AdviceRobo’s solution and collects all (anonymised) customer data, calculates risk scores and customer profiles and delivers this back to our clients. All data and delivery are connected through API’s.

The AdviceRobo platform has two API applications. The first is the Scoring API. The Scoring API facilitates the process of the consumer filling in the questionnaire. The second is the Reporting API. The Reporting API provides the customers of AdviceRobo with access to their data and settings.

Data is stored in an Azure MySQL database. This database is only accessible from a set of whitelisted ip addresses.

The technical quality of the Platform is measured by SIG (www.sig.eu) and AdviceRobo was granted with a Maintainability score of a solid 4,1 stars out of 5. The SIG approach and methodology is fact based. The approach is based on ISO IEC 25010 for good software development. Her benchmarking is certified by TÜViT ISO IEC 17025 and calibrated yearly.

We use cookies to help our site work and to understand how you use it. Click accept to help us keep making improvements.