December 26, 2018 | Written by: Kurt MacAlpine
Categorized: AI/Watson | Financial Markets
Share this post:
Growth depends on the right message
As an asset management company, WisdomTree is in the business of designing and distributing investment products. To sell our products, we work through intermediaries—financial advisors, retail wealth platforms and institutional investors—rather than selling directly to consumers. Though this model works great from a distribution efficiency perspective, it comes with the inherent disadvantage of making us one step removed from the final decision-making process.
Why does this matter? It’s all about growth. For companies like us to expand our base of investors, we need to target the financial professionals who evaluate and buy investment products for regular consumers. There are approximately 260,000 financial advisors in the U.S., and it’s the job of our relatively small distribution team to pitch them. We realized that the only way to manage this daunting ratio was through optimization—via prioritization. Though we had a lot of data on customer interactions, we lacked a roadmap to guide our team on which advisors to target, how to best reach them and what specifically to offer them.
Learning what works through AI
We closed this gap by working with IBM Watson Services to build an AI-powered lead prioritization system. It uses a machine learning model—based on the IBM Watson Personality Insights service—that sets priorities for our distribution team to follow. The model’s basic principle is that patterns of past interactions yield reliable insights into what works best. These records of interaction include all the conversations that the distribution team has had with advisors, including whom the conversation was with, what was discussed and the mode of contact. In each case, the quality of the interaction is rated within the model along a scale from positive to negative. When applied to thousands of interactions, the predictive patterns emerge.
After starting with our CRM system, we enhanced the model by adding marketing data alongside it. This has given our distribution team a more precise read into who they should be talking to and what the messages should be. And it continuously learns over time. Based on how each interaction plays out, our distribution team members score it: Was it a positive interaction? Was it a negative interaction? Was it an advisor that we should be talking to with the right message? Was it the right advisor but the wrong message? That constant training refines the model’s pattern understanding and makes it a more valuable tool.
Tighter processes spur growth
Our AI solution has shaped our distribution outreach efforts into a tight and efficient process. Our ability to prioritize which leads to focus on not only has increased efficiency, but has helped our distribution team to identify the optimal messaging. This combination of better targeting and message optimization ultimately has led us to engage more effectively with investors.
Watch Kurt MacAlpine talk about how WisdomTree worked with IBM Watson to create a machine learning model for their distribution outreach: