Knowing what the customer needs has always been a desired goal of most companies, not only in banking. They want to get more information and, if possible, replace general campaigns with better targeted marketing campaigns to smaller groups of customers with similar needs. Current marketing options allow the use of so-called “event driven marketing”, where the impetus for addressing the customer are implemented or not by non-specific events. If we actually know what life situation the customer can be found in and conveniently schedule a timely campaign, it can help us tremendously in selling our products and services.

In the introduction, we described a concept, which in its essence is not unfamiliar. The professional sphere has known about it for some time and even has implemented it somewhere on the basis of structured data. So what is new? Using unstructured data associated with the development of Big Data is gradually coming into prominence. Technology is not only financially accessible (you can get an entire solution for millions), but complex data processing also helps. Textual analysis of the Czech language or tools for converting speech, which implies there are no longer any utopias.

Just for clarification. One of the fundamental objectives of text analytics is to search for specific information in the text and convert them into a structured form. Therefore, in banking, we search for information in the descriptions of Internet banking transactions, notes from bankers when dealing with clients at the branches, emails or automatically transferred calls from call centers. Perhaps, in the near future we will see the possibilities of transferring information from voice recordings directly into forms. At least it is being worked on.

How did the bank successfully approach parents and holidaymakers?

We recently implemented a contract for one domestic bank. Our goal was to find new and as yet un-extracted information about clients and reflect it by using specifically tailored banking products. We worked with three types of data: notes of personal bankers, descriptions of commands in Internet banking and identifiers of transactions on payment cards.

There were a total of two target groups identified. The first and smaller group was composed of parents of children who found themselves in financial distress in September. An analysis of key words was used to identify them, as the bank did not have a specific existing product for children for these clients, as they had no idea about them having children. The second group consisted of holidaymakers, who were people who had paid for a trip, ticket or bought travel insurance. These clients were offered a credit card for unplanned expenses as a financial reserve.

The result? The overall conversion was 6–7% in the end, which meant it had doubled compared to the classic combination of direct mail and calls from the call center, even though  the sales potential for half of these groups was estimated at (gained by classical models) about 5 percent. The Bank was able to increase sales by more than tenfold.

Thousands of new clients per month

Full implementation of the transfer of data from an unstructured to structured form and identifying target groups took place during a single month. The system is not able to regularly identify those thousands of clients each month and at the event-driven marketing just at the right timing is absolutely key to the success of a campaign.

Even though text analytics are able to identify a relatively small number of cases, it does it continuously according to specified intervals, which is certainly worthwhile when the conversion rate is tenfold. Additionally, this solution can be relatively easy to modify and extend to areas other than marketing activities. For example, they have the potential for the field of risk or extracting can help significantly when assigning tasks to a call centre.


Author of the article: Dagmar Bínová, Business Consultant at Adastra

Source: Marketing Sales Media (11/2015)