- use case -
Automation of interim statements at Bank Van Breda
Learn how Bank Van Breda is exploring the automation of data extraction from interim statements that has a potential to save 333 hours per month.
Bank Van Breda is a Belgian bank specialized in servicing entrepreneurs and liberal professions, both in their private as professional challenges. We sat together with Nanou Van Deuren who’s been working at BVB for 12 years now. She’s team leader of the 7-headed credit analyst department for liberal professionals.
The challenge
When you’re working in a credit department, you know the drill: documents, documents, documents, … It’s a key role in the job of a credit analyst and the information on these documents are crucial for further processing credit applications. Nanou her team at Bank van Breda is responsible for analyzing credit applications for liberal professionals. To build a good case and do an extensive analysis of the client, her team often needs interim statements (a report of financial results covering a period of less than one year) from the account.
Interim statements are crucial in the credit analysis process, because they convey the performance of a company before the end of normal full-year financial reporting cycles. These reports can also alert her analysts to recent changes that meaningfully affect the corporation.
The problem with interim statements
While annual reporting has standardized formats for submission, that’s not the case with interim statements. Every account has its own way of doing this reporting, which makes it a challenge for credit analysts to work with.
“Depending on how it’s formatted, the pre-processing of the documents can already take up to 30 minutes per interim report. This means: just analyzing the structure of the documents and in certain cases we even have to calculate balances ourselves. If you know each credit analyst processes up to 5 cases per day, this means 2,5 hours per day of pre-processing documents.”
Furthermore, the information on these interim statements and the balances need to be entered into our internal systems. That was a job that used to be processed manually by their credit analysts. It’s not a fun job, it takes a lot of time and is error-prone.
On the lookout for a solution
“We’ve been looking into potential partners to help us optimize this process, since it would take up too much time to develop this system internally. Together with Metamaze, we’ve kicked off a project to start automating this process.”
The scope of the project
To optimize this process, Bank van Breda tested with Metamaze to automatically extract information from these interim financial reports and to make sure this information gets into their internal systems.
On average, these interim statements consist of both a interim balance sheet as well as interim profit & loss sheet. A lot of information on these documents are interesting and needs to be extracted, like: stock, inventory, accounts receivables, bank account, working capital, long-term debt, short-term debt, …
Potential results of the project
When you know that Nanou and her team at Bank Van Breda processes up to 8000 credit applications per year. There is a potential of 333 hours per month to be saved in her department only.
The model has been trained and has an accuracy of 87% based on the learnings of only 290 documents. In these documents, up to 76 entities are extracted.
“We decided to work on this project with Metamaze because we’ve experienced a good collaboration on past projects. The proximity of their contact persons and the CEO is an important aspect for us in a collaboration. We always have the feeling that Metamaze takes the time to answer questions in-depth. They think with us and are transparent in their communications. Timing, deadlines, scope project, … were respected at any time.”
