Using machine learning to help small businesses

**Scenario **

We route and record calls between the business and customer, which gives us a wealth of data. This feature was our first to apply machine learning to these calls and determine whether the customer made an appointment or not. For a small business owner, this removes the burden of tracking call outcomes. It also opens up a lot of opportunity to further analyze calls and see what leads to success.

Approach We had to be careful with the way we presented this, because it is a phone call between a business and their customer. For many users, this would be their first experience with a product openly describing how machine learning uses their information. I iterated on this language quite a bit, with extensive feedback from legal and senior executive level feedback on terms like “machine learning.”

We conducted research with several dozen participants, changing the copy as we went. I was consistently surprised by people’s reactions, which were about 20 seconds of disbelief (“A robot listens to my calls?!”) followed by acceptance (“Well that’s neat if it helps me!”). We were delighted to discover that businesses wanted to know the basics of how it worked and how it benefited them, and we not intimidated.

Results

This project led me to join a cross-functional AI group which produced Google’s first Human Centered Artificial Intelligence cookbook. I was one of two content strategists for the UX section.