The best deposits price optimization solutions are decision-support solutions, in that they help humans make better decisions. It is only within recent years that advanced technology—such as predictive analytics and big data—has been available in a way that deposits manager could easily apply them to optimize their rates and hit their balance and profitability goals. Coming out of the long, flat, near-zero rate environment we have been in, most big banks, and a handful of large and regionals have tools in place and are ready to pounce as rates start to tick up. I am often asked what to look for when shopping for this sort of technology, and offer the following as a guide:
Delve deeply into the models. They are not all created equally.
Look at sample what-if analyses and make sure that what you see reflects what would be of value to you in your job later this year as rates are climbing.
Talk about data internally and with your solution provider. How many separate databases will you need to extract from? What security measures do you need to ensure? Who will normalize the data? How long will it take to get the data into usable format? Processing your data can be a key success/failure point for your project. Find out about data ingestion, big data pipelines, etc.
Build high-definition pricing models using account-level price sensitivity analysis. Chances are that you won’t price at the account level, but your data should probably be analyzed at that level so roll-up analyses are as accurate as they can be.
See a real-data demo of the optimization workflow. True constraint-based efficient-frontier optimization that balances tradeoffs between balance growth and profits will deliver superior results compared to older approaches.
See and understand interactive flow-of-funds forecasting and reporting so that you will be able to manage cannibalization and attrition.
Make sure that your in-house team will be able to use their own models if you choose to go that route. Talk to the people who build models to see what advantages they can point out about their approaches. Yours might still be best, but their perspective is based on a broad swath of customers and they often have very valuable insights.
Find out how often you will be able to update your models or coefficients based on updated data. Monthly is probably the minimum you’d want and more and more banks are moving to weekly. Pedigree in big data is key here. Ask about your solution provider's technology stack and try to talk to the data scientists. They are a hidden linchpin in the success of your project.
Model recalibration should be a major part of your discussions—especially given that your first models will be built based on a long period of low signal-to-noise ratio, and as soon as rates start top move frequent recalibration will be critical.
Find out what other data sources will be used, including macroeconomic data and competitor rates as well.