It is core business for all banks to provide sound financial advice to clients. Competitive advantage is gained by providing better advice than competing financial institutions.
The recommendation by the financial advisor at the bank to the client is provided as an advice document which includes the best possible financial advice for the client given his or her current financial situation.
In this particular bank, 20% of all recommendations are reviewed by other expert advisors. Reviewing is important for the bank since it ensures high quality of their recommendations which in turn has an impact not only on the client's economy but on the profit margin of the bank as well.
The expert advisor role is to:
- ensure that the advice provided is of high quality
- educate the first advisor by giving feedback so he or she can improve her recommendations in the future
- detect and correct flaws in the recommendations
Striving for further improvements in the financial advice provided by the bank to their clients, Findwise was tasked to:
- increase the number of recommendations being reviewed
- improve the overall quality of financial advice given to clients
- increase the competence of the whole advisory team
Findwise implemented a Watson Explorer-based, automated review process for financial advise.
Watson’s job is to analyze the recommendations made and detect those recommendations that should be reviewed by an expert advisor. In this way, the expert advisors are more likely to review those recommendations that actually contain flaws instead of spending time on reviewing advice that is correct.
The automatic analysis is done using a combination of Natural Language Processing (NLP) and a metadata model based on rules crafted by a senior reviewer. The model identifies the recommendations with the lowest quality and selects them for manual review. In order to continuously, and automatically, improve a ML model for constantly refined detection of flaws in recommendations the selected and manually corrected, recommendations are also internally tagged and used as base for machine learning (ML),
In addition, recommendations are used to give feedback to the advisors, improving the quality of the advice being given by, and the competence of, the whole advisory team
Adding Watson to the reviewing team has resulted in a greater number of reviewed documents, higher quality of the recommendations provided to clients and improved internal knowledge sharing among the finacial advisors. By collecting, tagging and storing the advice and the verdicts, both from the reviewer and Watson, more advanced learning models can be built.
With the ability to measure quality of the analysis the bank now has a great platform to build advanced reviewing capability of financial recommendations in real-time. And clients get better financial advice, which gives the bank competitive advantage - a win-win for both the clients and the bank.