Our outbound dialing scenario automatically calls a customer, and if the customer answers the call, the customer is connected to an available customer service agent. This design eliminates dialing and waiting times, handles busy situations, and replaces fax and answering machines—all while staying in compliance and maintaining a high predictive connection rate. Agents are involved in a call only when live customers answer. But how does the dialer know when an answering machine has answered the call versus a live person?
AMD and Algorithms
Answering machines and live people answer a call in different ways. A live person will say, “Hello?” or “Hi,” and silently wait for a response, whereas an answering machine will respond immediately with a long sentence followed by an option to leave a message. Detection algorithms might show, for example, that if a signal lasts over 3 seconds, it is an answering machine. Otherwise, it is a live person. We use a component known as an Answering Machine Detector (AMD), along with detection algorithms and Convolutional Neural Networks (CNN) for speech and image processing and to determine who or what has answered the phone.
Agents Are Connected Only When They Have to Be
When you know what has accepted a call, you also know when to leave messages and when to connect live callers to agents. If it’s an answering machine, your contact center’s dialer can leave a prerecorded message or have an agent record a new message. If it’s a live person, however, the outbound dialing scenario will connect the call to an agent. This saves agents’ time because they only speak to live callers and do not spend their time leaving messages on machines.
- Outbound dialing scenario automatically calls a customer
- Answering Machine Detector component learns to leave a message on a machine or to connect a live caller to an agent
- AMD saves time for agents
- Contact centers increase the probability of connecting to live callers
We are happy to answer questions and discuss specific workflows and requirements on a live presentation online or set up a pilot project to trial the solution without paying any license fees