How supply and demand come together

Older generations remember it like this: you’d go to the bank and your trusted advisor would give you tips on how to invest your money. Modern banking works differently, but it still relies on recommendations that spark a customer’s interest and ultimately lead to a deal. With this in mind, a research collaboration was launched between FH Upper Austria in Hagenberg and the software developers at bluesource.

Anyone who regularly browses the web and shops online is constantly feeding recommendation systems: intelligent algorithms that analyze user behavior and, based on that, suggest a wide range of products, trips, and much more. These virtual advisors are also becoming increasingly important in the banking and insurance sectors.

At FH Upper Austria’s Hagenberg Campus, a team led by Professor Ulrich Bodenhofer is working on such a recommendation system for the company bluesource. The company specializes in app development, employs around 40 people, and operates not only its headquarters in the Hagenberg Software Park but also additional locations in Linz and Vienna. One of bluesource’s products is the ‘mobile‑pocket’ app, which digitally stores all your customer loyalty cards so you don’t have to carry them in your wallet. Beyond that, the company provides this card‑storage functionality for around 40 banking wallets across Austria.

Relevant Recommendations
Together with researchers at FH Upper Austria, the goal is to lay the foundation for future products—specifically, new apps. Ulrich Bodenhofer, AI specialist and head of the Artificial Intelligence Solutions degree program, explains:
“We are developing algorithms for a tool, commissioned by bluesource, that predicts users’ investment behavior based on their interests and generates recommendations accordingly.”
This means users will receive suggestions for financial products that align as closely as possible with their preferences.

To achieve this, the data from the customer loyalty cards stored in the banking app are matched with data from financial products.
“All of this is done in compliance with legal regulations such as the GDPR and the EU AI Act. All data is fully anonymous — we don’t know users’ age, gender, place of residence, or ethnicity,” says Thomas Otzasek, Chief Data Officer at bluesource.

Every card tells a story
Specifically, the researchers look at which loyalty cards a person has, when and how often they use them, and which ads they click on in the app. From these parameters, it's possible to infer what someone is interested in and what kind of “life world” they belong to. “Every card tells a story,” says Ulrich Bodenhofer.

For example: if someone has loyalty cards from a hardware store, a furniture chain, a toy shop, and the Upper Austria family card — and uses them regularly — one might infer they’re part of a young family. In this case, the algorithm might suggest products like a building savings plan or a home loan. If the cards and their usage patterns point to an older, financially well‑off user, the system would likely recommend securities or private retirement plans instead.

Different Life Worlds
Thomas Otzasek explains: “The concept of ‘life worlds’ isn’t only about which products are recommended. It’s also about how customers are approached — how the recommendations are written and designed visually. That’s something that would look different for each of us.”
Within the app, the various banking and insurance products will also be described and explained. “This is how we contribute to the financial education of our users,” Otzasek adds.

Ulrich Bodenhofer points out a unique feature of the planned app:
“In the solution we’re working on, user profiles are not hidden. Users can check what information is stored about them. Everything can be reviewed and, if needed, changed.”

The collaboration project is scheduled to run for three years and will conclude at the end of 2025.

(AI-generated translation)