Zoe.ai logo
Schedule a demo
This website uses necessary and analytic cookies to measure traffic.
We may place these for analysis of our visitor data, to improve our website and to give you a great website experience. For more information about the cookies we use manage the settings.

AI-based credit scoring

  • Michal Krňák
    Michal Krňák
  • 1 min read
AI-based credit scoring
Table of Contents
  • The problem
  • User behavioral detection

In my last post dedicated to using cases in insurance and how Zoe.ai can help to strengthen the investigation process, I mentioned 10% of lying users while filling in an online form. This problem concerns an online loan application process too.

The problem

There are millions of users all over the world, who are inconvenient for one of many criteria in – mostly very rigid – loan approval scoring process. And there are also banks and other loan providers with huge losses of profits for these ones.

Imagine there exists some 10% of applicants who are completely trustworthy but finally rejected by loan companies. And what if I tell you behavioral data could help their final approval with no incremental risk?

User behavioral detection

So let’s take a closer look at our user behavioral detection.

Every single visitor of a web page or an online form has his own significant and unique key metrics, which can help to identify him precisely in an online environment every single time. Zoe.ai is able to learn about these key metrics of – let’s say – non-problematic, non-suspicious users and applicate its artificial intelligence to all the next ones with similar behavior matching over the specified threshold.

We are talking here mainly about metrics such as “flight time” - how long time you spend by filling in a particular column, keystroke dynamics, mouse moves, if and how many times you move with virtual slider, how long does it take you to read terms and conditions and many more. There exist a significant connection between statistics from particular metrics and non-problematic users. These users have then scored accurately compared to the relevant population with very similar biometrics. After that, user’s personal data are usually verified through third-party registers. Many of them are finally “on the edge” – is usually means tightly rejected.

No alt text provided for this image Thanks to Zoe.ai, there is another point of view on those users, who can be (and in the vast majority really are) reliable customers but do not get a chance to prove.

According to our findings, revenue connected with behavioral biometrics used within the loan approval process raised up to 2%.

Loan approved

So what are you waiting for? 😊

Most frequent use cases our Zoe.AI tool addressesHow to identify suspicious aspects of online behavior

Did you have fun reading it?

Let's talk about it more!

Prague

Aguan s.r.o.Kaprova 42/14110 00 PrahaCzech republicIN: 24173681+420 222 253 015info@lundegaard.eu

Prague

Lundegaard a.s.Futurama Business ParkSokolovská 651/136a186 00 Praha 8 - KarlínCzech republicIN: 25687221+420 222 253 015info@lundegaard.eu

Brno

Lundegaard a.s.Ponávka Business CentreŠkrobárenská 502/1617 00 Brno - jihCzech republicIN: 25687221+420 222 253 025info@lundegaard.eu

Hradec Králové

Lundegaard a.s.Velké náměstí 1/3500 03 Hradec KrálovéCzech republicIN: 25687221+420 222 253 015office.hradec-kralove@lundegaard.eu

Zoe.ai logo
Fully individualized experience through AI Deep recommendation.
For developers
Deep recommendation
Products
Contact us

All rights reserved by Lundegaard a.s.

Services provided by Aguan s.r.o.