PwC worked with an asset manager to define what behaviors lead to successful sales outcomes to increase the effectiveness of the interactions between Sales Employees and Financial Advisors. The asset manager wanted to better understand the potential value of applying machine learning tools to extract insight from the recordings of the sales employee/financial advisor phone calls to improve sales effectiveness.
- Created a master dataset from the client’s disparate systems (e.g., audio recordings of calls, CRM, Transactions).
- Used commercially available speech-to-text APIs and proprietary deep learning models to extract useful features from tens of thousands of call transcriptions and call metadata.
- Combined this information with structured transaction data to gather additional insights.
- Identified features that were correlated with successful sales outcomes, such as the number of exchanges in a conversation, number of hesitations, topics discussed, and whether the call went to voicemail.
- Identified pre-call factors that increase the likelihood of success (e.g. interaction history, timing of call, product traded, etc.)
Impact on client’s business
- Developed recommendations based on the significant features, such as the optimal time to call clients to maximize the likelihood that they answer the call.
- Exploring more granular segmentations of sales employees and financial advisors and prescriptive recommendations for interactions (e.g. who, what, when, etc.)
- Several of the findings, including suggested interaction frequencies, channel patterns, and improvements to sales scripts were incorporated to improve sales training programs.