Machine Learning Producing Profits for Call Centers

Machine learning is an explosive new trend in business.  By discovering patterns in large streams of data at near-instantaneous speed, machine learning can now drive a car, translate words into a foreign language while you speak using what sounds like your own “voice”, accurately diagnose disease, and even perform years of scientific research in a matter of hours.  

The possibilities seem almost magical. The practical results are already stunning. And the business world has started to pay close attention.  Venture investments in machine learning startups quadrupled last year to $300 million. This is a hot technology. But few people have asked yet — what could machine learning do for call centers?  

When I studied math in college I had no idea I would end up creating software. My career began with a South African hedge fund, designing algorithms to trade stocks.  The software I built proved very profitable.  We beat the market by discovering patterns that had not been observed before.  After building a successful program for my employer, I found myself with just enough free time to think about starting a new company.  So I began looking for new applications outside of finance...  And that’s how DataProphet was born.

Life is full of discoveries. To my own surprise, in 2014 we began focusing on the call center space. The interesting thing about outbound call centers is that they have large lead lists, and a ton of information on historical performance. There is a lot of data to work with. No one that we knew at the time was thinking about machine learning for call centers…  Lots of attributes were being measured and used for call center management. Most were and still are using traditional statistical models. But few people understood the dramatic potential that still remained unobserved, hidden in all that data.

DataProphet creates mathematical models and algorithms to find correlations between measurable attributes of individual agents, customer leads, and sales results. I cannot speak for the other companies in the space, but we’ve seen impressive results to-date. By learning through our software what really makes a bottom line difference, we’ve increased call center sales by up to 36%!  

Done right, implementation requires very little to no operational change. Machine learning relies primarily on access to historical data. Results can be improved significantly by simply reshuffling who gets assigned to what lead, for example. We call this Agent Lead Matching (patent pending). One of the most powerful things about machine learning is that results are scientifically measurable using control groups. Not only do results get better, but you can prove it!

Never before in the history of this industry have gains of this size been realizable with such little effort.  The applied mathematics of machine learning are going to create tremendous new benefits out of the mountains of data that human organizations create. And I think nowhere is there more opportunity than in the call center industry.