The exponential progress of artificial intelligence (AI) opens up new ways to bind customers to a company or brand through excellent service. Not only for providers of consumer goods are the “smarter” digital employees from generation to generation becoming steadily more interesting, but also for financial service providers: customers from all sectors expect personalised, error-free answers in seconds. Currently, service, whether on the phone or in the branch, is too slow and too expensive for them. Opening an account, for example, takes too long, and the annoyance about fees and minimum deposits has increased again since the turnaround in interest rates.
Those who are content to jump on the moving AI train in order not to lose touch will soon be caught up in the competitive pressure. In order to secure or even expand their own market position in the long term, financial providers are well advised to leverage optimisation and innovation potential, for example with the increasingly popular chatbots.
Corona and AI as a gamechanger
During the Corona pandemic, banks and their customers had to conduct all business at a distance for the first time – in addition to regular account or deposit management, also advice-intensive services such as loans. Financial providers can make a virtue out of this hardship by automatically answering questions that constantly arise around the clock. From the customer’s point of view, too, a chatbot has more to offer than a Q&A, no matter how extensive, but always monologue. That is why more and more banks are developing their own chatbots for websites or apps as alternatives to ChatGPT.
But chatbots can do more than tap efficiency reserves and better inform customers. Their interaction with the audience also provides a wealth of data on interests and buying behaviour. Based on such insights, financial service providers tailor the advice given to individual customers just as precisely as they tailor the product portfolio. And not just sporadically, but continuously: because the AI algorithms learn iteratively from the interaction, they know more and more precisely what the customers want and follow the change in demand. The system does this on its own. Apart from regular monitoring and maintenance, there is no need for any further investment.
Ein professionell gestalteter Chatbot beantwortet Fragen schneller und bequemer als ein seitenlanges Q&A, das die Interessentin aktiv durchforsten muss – ohne Gewähr, dass sie finden wird, was sie sucht. Anders als der Sachbearbeiter am Telefon oder die Beraterin in der Filiale ist der Bot auch außerhalb der Kernarbeitszeit ansprechbar. Zudem lässt er sich skalieren. So trägt er einerseits zur Bindung des Kundenstamms bei, anderseits verschafft er der Bank oder Versicherung einen Wettbewerbsvorteil in der Akquise.
Als digitaler Kollege, Datenquelle und Lernhilfe steht der Chatbot auch dem Personal der Bank oder Versicherung zur Seite. Kundenanliegen, die er nicht abschließend bearbeiten kann, leitet er an den zuständigen Mitarbeiter weiter. Je nach Verlauf des Dialogs sucht er im Hintergrund Produktinfos heraus, die der Mitarbeiter der Kundin anbieten kann, oder weist auf Sicherheitsregeln hin. Überdies helfen Chatbots bei der Bewältigung des Fachkräftemangels.
Entwickler branchenspezifischer KI-Apps weiten deren Funktionsspektrum zügig aus. Bald wird es möglich sein, per Chatbot den Kontostand abzufragen, Geld zu überweisen oder den Verlust der Kreditkarte zu melden.
What does a chatbot have to look like so that users are satisfied with it? What do they particularly value? In addition to hard performance criteria such as response time, scope and relevance of the integrated functions, customers appreciate chatbots that communicate as humanly as possible. Especially because the interaction with the bank is a serious matter, the bot should “think”, be attentive and empathetic.
The bot delivers added value if it does not fob customers off with pre-formulated blanket answers, but recognises individual needs or interests such as impending late payments or savings goals in the dialogue and responds to them with customised tips or offers. This benefits customers and staff alike.
However, data protection must not be neglected in the focus on what is technically feasible. Chatbots are also subject to the obligation to inform according to the GDPR, which means that the user must inform customers about the collection and use of their data before each interaction with the bot. The collection of personal data in particular requires explicit consent. The collection must be limited to such information that is functionally necessary in the currently active context. With legally and technically correct design, the data protection risk of a chatbot corresponds to that of other apps that process customer data.
Solution from Consileon
The development of a customised chatbot is quite an elaborate IT project. The specification of the goals and the scope of functions may seem manageable, but it can escalate when maximum accuracy of fit and differentiation are required. However, the initial machine learning process, the so-called training of the algorithm, takes the longest. It also requires a large amount of data. With a broad customer base, larger financial providers in particular have a good starting point.
For a quick start in the use of a chatbot, Consileon offers a neutral basic framework that can be easily integrated into the user’s system landscape. Our MyPersonalGPT is based on ChatGPT 4.0. Banks and insurers can expand the bot with master data, documents and expertise into an industry- and company-specific multitool. Among other things, MyPersonalGPT helps clerks and advisors to maintain customer contacts, write texts, conduct research and interpret business reports. By carefully selecting the sources the bot accesses, financial service providers provide their staff with information they can blindly rely on.