An effective data management strategy is crucial
Modern data management and explainable artificial intelligence (XAI) are no longer negotiable components for banks and insurers. In this interview, Tom Hartung (InterSystems) emphasizes the challenges posed by heterogeneous IT landscapes and legacy IT.
Mr. Hartung, the pressure on traditional banks to compete and modernize is enormous. Is this a result of increased expectations arising from modern apps and GAFAs?
Tom Hartung: In my opinion, the increasing pressure to modernize is primarily related to the legacy IT architectures that we still see in many traditional banks. These make it difficult to access data in real-time and create data silos that prevent the systematic use of information. In addition, there is a staff shortage. IT experts who are still familiar with the old systems and who understand the relevant programming languages have become rare – with many retiring. But of course, the pressure for change on established financial institutions is also caused by the so-called neobanks. With their low-cost services and user-friendly apps, they can appeal to younger target groups.
It is said that bank management is central to the modernization of financial institutions. But wouldn't it be much more important to make the customer front-ends more appealing?
Of course, an attractive customer front-end is important, both to retain existing customers and to attract new, younger customers. But bank management is existential – it is the only way a bank can survive in the long term. It is therefore crucial that key aspects such as the financial situation, the customer churn rate or the risk factors can be quickly retrieved from the data at any time. In order to maintain the right balance between profit and risk, the right data must be immediately available - for example, to determine whether the bank will remain financially stable if there’s a certain percentage of customer churn.
In your opinion, what is the major problem that financial institutions currently face from an IT perspective?
The main problem is the very fragmented IT landscape with disparate data sources. In many cases, this is due to acquisitions, organic growth and the continued operation of legacy systems. Such systems foster incompatibilities and the creation of data silos. In many financial institutions, data is distributed across different departments. This makes it almost impossible to compile the relevant information at short notice. A risk assessment, for example, can take several weeks. By then, a large proportion of the results will be out of date. Another obstacle is the lack of a proper database. This makes searching for, identifying and processing information extremely time-consuming. Even the best AI algorithms cannot generate useful results from incomplete and unreliable data.
So, what benefits do you see from better data management?
Effective data management provides the bank with a comprehensive and holistic overview of its entire economic and customer situation. Financial institutions will also benefit from a reliable and accurate database - an important prerequisite for quickly creating trustworthy reports, well-founded analyses and the successful use of AI.
Are there any risks associated with a modern IT financial architecture that focuses on data management?
On the way to a data-driven company, it's all about creating added value from the available data. For example, by analyzing this data. One risk associated with this approach is that data analyses are incorrect if they are based on incorrect data models. For example, an inaccurately trained prediction model will not provide accurate forecasts. To reduce such risks, choosing the right partner is crucial. The solution provider should have been in the market for a while and therefore be well experienced. However, it is also important that the partner's business model is based on long-term customer relationships. This is because smart and better data management is a marathon, not a sprint. You also need a good pacesetter.
Which advantages does a solid bank management offer, for example to make regulatory reporting easier to handle?
If bank management is based on a solid data platform, data can be re-used and made available immediately. The bank can then respond faster to regulatory reporting requirements. But I do believe that the aspect of data monetization is much more important. For example, AI-supported bank management can be used to identify typical patterns of customer churn. This allows suitable countermeasures to be taken quickly.
And where should financial institutions start to turn their data into added value?
Financial institutions should ensure that they have an appropriate data culture. This means that every employee should have a clear understanding of the data they are responsible for. They should know how this data is defined and how and at what intervals it should be updated and maintained. A good awareness of the value that data represents in today's organizations is certainly helpful. The second important aspect is a reliable database that can serve as a basis for managing and analyzing data. However, this is still a challenge due to the frequently encountered "spaghetti architectures" with their heterogeneous formats and interfaces. If you want to avoid replacing the entire IT landscape, you should implement an intelligent end-to-end decision platform that wraps itself around the various data pools and connects them without affecting the existing infrastructure. This allows data to be extracted, normalized and shared with other applications as required, no matter the source. If this platform can then also orchestrate business processes and provide native analysis functions, the prerequisites for smart data usage are very good.
Many financial institutions hesitate to move their systems and data to the cloud, despite the obvious advantages. Couldn't data management in the cloud make more sense?
In general, data management in the cloud is highly recommended. But this requires a hybrid platform that integrates all relevant processes to ensure that no data is transferred. Because if a bank uses the analysis tools of its cloud provider, redundant data may be created. In addition, the transfer of the data to be analyzed and the evaluated data records between different systems, known as ETL (extract, transfer, load) processes, is particularly time-consuming and costly. The on-premise model, on the other hand, is recommended for GDPR-relevant data and applications such as payment transactions and lending. Hybrid deployment models, i.e. simultaneous operation of on-premise and cloud, are therefore ideal for financial institutions.
Considering the development of bank management and data management in the future (especially with regard to AI and XAI): What will the technology look like in five years' time?
At the moment, almost everyone is talking about artificial intelligence. But in many places, it is still unclear how and for what purpose it can be used. In addition, few banks have data that is structured and consolidated enough to be used for meaningful analyses. And it is likely to take some effort for most institutions to prepare the data in an appropriate way. In my opinion, XAI, i.e. Explainable Artificial Intelligence, will give users and customers the confidence of being able to look behind the analysis model. Explaining how the decision has been made by AI can hopefully also reduce or perhaps even eliminate the scepticism that AI is facing.
Mr. Hartung, thank you very much for the interview.
First publication of this interview: August 2023, IT-Finanzmagazin