Berkeley, CA, Dec. 11, 2020 — 6Sense, an AI predictive engine for B2B organizations to predict sales, has teamed up with a group of UC Berkeley students to design a predictive model for determining the best time to call prospective leads. The predictive model is integrated into a Predictive Call Scheduler tool that helps telemarketers create the optimal schedule for calling each lead as well as generate topics of interest for each lead.
According to Moore’s Law, technology is advancing at an exponential rate. But for some reason, telemarketers are still using methods from the 1900s to reach potential customers. According to sales statistics on cold calling, the success rate of a cold call is around 2%. To put this into perspective, for every 100 cold calls a sales representative makes, 98 of these calls will be unanswered. If a company hires 20 sales representatives who each makes 100 calls per day, only 40 of these calls will be answered.
6Sense and the UC Berkeley team believes that cold calling should be dead, and every business should be evolving from this outdated method of reaching prospective customers. They want to emphasize using the tools of the 21st century to help businesses reach their customer base and cater calls to individuals who will actually answer the call.
After an entire semester of research, analysis, trial, and error, the UC Berkeley team has developed a model that predicts the probability of a successful call based on a lead’s activity and shared personal data such as job level, company, and industry. The team utilized supervised machine learning algorithms, optimization methods, and designed many impactful features that contributed to improving the precision and recall of the model. The model is integrated into a pipeline where real-time and stored data is automatically, cleaned, transformed, and inputted into the model for predictions. The input to the model is a list of contacts to call. The output of the model is a success probability for each contact from the inputted contacts list as well as conversational topics tailored to each contact’s interests. The system then sorts the outputs by success probability and job level of the contacts from high to low where higher job levels such as directors will be prioritized over staff. The current release of the product currently has the ability to schedule calls for any specific date or schedule calls across an entire week.
This product’s launch will allow 6Sense and other companies to accurately schedule when to call a prospective lead and suggest potential topics of interest to guide the conversion. More successful calls will ultimately generate more opportunities and lead to more sales for the company.
Project By: Kelly Huang, Yinglu Deng, Sunny Sun, Winnie Lu, Jin Wu, Jenna Yu
Github Repo: https://github.com/kellyhuang21/6SenseCallPrediction