This Call May be Monitored to Help Us Help You
Can algorithms really make us more empathetic? Cogito is using machine learning, data, and nudges to keep us all from pulling our hair out when contacting customer support. Co-founder Joshua Feast: “in some ways, we think of ourselves as a cyborg company, helping humans be their best selves.”
Everyone knows the utterly miserable experience of having to call a 1-800 number for customer support. You’re frustrated, they’re frustrated, everyone wants to leave the call with a resolution, but no one enjoys the experience. Boston based start-up Cogito is using data to try and help us all! With the growing importance of customer satisfaction in an age where consumers can take to social media to complain loudly about bad experiences, it’s increasingly important to get it right the first time.
Cogito’s solution is to marry behavioral science, real-time and historical data, call context, and industry and company norms to assess the call and recommend better actions for call center representatives and build customer empathy. The result is a company purported 20% improvement in customer satisfaction. On the call, Cogito tracks real-time metrics such as “dialog speed, pauses, interruptions, volume changes and tone,” and compares with call context, successful historical calls, and industry and company norms to provide actionable conversational recommendations and behavioral cues to call center representatives. Cogito has already attracted clients such as Humana and BlueCross BlueShield, Aetna. Results so far have been encouraging – with customers experiencing and average call time decline of 10% and fewer average number of call backs.
Call centers are an ideal lab for data collection and training the machine-learning powered algorithms. While the fully automated call center may be a ways into the future (who would want to try and have Siri solve all our problems?!), Cogito’s program is helping the call center works become more efficient and empathetic on calls – by recognizing key words, trace emotions, and energy levels.
Cogito has the Pentagon to thank for their machine-learning software’s ability to recognize distress and emotion in customer calls. The company’s early genesis in the MIT Media Lab granted the company access to Pentagon-funded research that served as the perfect training data. The data, which came from a project to help veterans with PTSD, was audio files from interviews with patients with psychological problems, annotated to capture changes in emotional state. From there “Cogito applied deep-learning algorithms like those behind the improved speech recognition in assistants like Alexa to the Pentagon’s data, and audio from call centers and other sources.”
While Cogito is helping their clients’ call centers become more effective today, the future of this technology is still to be seen. A couple of open questions for the future:
- How technology like this can be designed to work across different groups (social, economic, cultural, IQ EQ or educational level) equally well to avoid potential difference in service across groups
- If customers will accept the extra levels of analysis being applied to the calls – which currently does not have to be disclosed
- How to ensure that the technology continues to be used in a responsible way – keeping it out of the hands of those who may want to use behavioral nudges to convince a customer (or voter!) instead of just helping them
Student comments on This Call May be Monitored to Help Us Help You
I think that the area of natural language processing and quantifying interactions is an extremely rich vein for data-driven companies. In another class we were exposed to a similar product called Quantified Communications that is attempting to help coach executives to be more better public speakers. It sounds like the base of this product could be leveraged in a similar coaching way in order to further improve call-center interactions and make customers happier.