How Genesys Cloud delivers outbound calls to agents
Genesys Cloud follows this process from the time an outbound campaign places a call to the time an agent answers the call.
- An outbound campaign places a call.
Call progress analysis begins:
- Signaling analysis begins.
- Network tone analysis begins.
- An entity, for example, an IVR, answering machine, fax machine, or live person, answers the call.
On enabling post-connect call analysis, the all call analysis begins.
Live speech detection (LSD) begins.
Live person detection begins.
Answering machine detection begins.
On enabling answering machine detection (AMD), call media analysis begins.
- Live speech detection (LSD) begins.
- Live person detection begins.
- Answering machine detection begins.
- Call progress analysis classifies the called entity.
- Call progress analysis and on enabling answering machine detection the call media analysis completes.
- Media microservices or premises-based Edge publishes the call information.
- The system places the call in an ACD queue.
- ACD logic routes the call to an agent with the correct skills.
- The agent answers the call.
- The agent and the caller communicate.
Tips for improving delivery time to agents
To reduce the time for Genesys Cloud to deliver outbound calls to agents, administrators, and contact center managers, perform the following steps:
- Disable post-connect call analysis. This action removes all the call analysis steps from the process after the call connects and also disables answering machine detection. Call progress analysis still classifies the called entity but does not distinguish between a live voice and an answering machine. To disable post-connect call analysis or to change the selection, see Create a call analysis response and Edit a call analysis response.
Call analysis classifies pre-connect events, such as Special Information Tone (SIT) messages and tones. However, when the line is connected, the transfer action is taken. This transfer action means that audio events that are expected to occur post-connect, such as live voice, machine, busy, and fax, are not detected, and applies to those indiscriminately. As a result, the Disable Answering Machine Detection box is automatically selected when Disable Post-Connect Call Analysis is selected, because AMD cannot be performed post-connect. This selection provides the fastest handling of live voice, but with the lowest accuracy, as all machines, busy, and faxes take the transfer action.
- Disable answering machine detection. This action removes the general speech mechanism of call analysis from the process, which would otherwise be used to distinguish between live voice and an answering machine. Call progress analysis classifies the called entity and can still classify an answering machine when a machine tone or audioprint is first encountered. To select Disable answering machine detection or to change the selection, see Create a call analysis response and Edit a call analysis response.
Note: If AMD is disabled, the system still accurately prevents around two-thirds of voicemail/answering machines from going to the live voice response action since many known default carrier messages can be identified almost immediately by way of spotting the voiceprint. Busy carrier announcements and tones, which typically account for about 3 percent of all calls, and faxes, are also detected and prevented from taking the live voice response action.There are three mechanisms for detecting machines:
- General speech, based on timings
- Tone, based on the machine beep
- Audio print, based on the Genesys audio fingerprint technology to detect common, pre-recorded carrier machine spoken messages, including busy and SIT.
Since the tone and audio print mechanisms are very accurate, only the first mechanism, general speech, is disabled when Disable answering machine detection is selected. This allows call analysis to not only continue to detect machines with very high accuracy for tone and audio prints, but also other tone and audio print events as well. This includes faxes, busy tones, and busy spoken messages. The general speech mechanism is not only used for machine detection but also for live voice as well. This means that call analysis does not take the full duration of time normally required with the general speech mechanism to try to distinguish between live voice and a machine. Or rather, after only about 0.3 seconds of speech, a live voice will be classified. For the typical short “hello,” live voice detection occurs about 1 second faster than standard answering machine detection.
The effective change in classification behavior when selecting Disable answering machine detection, is connected calls that start with personal answering machines, live voice greetings, and any carrier pre-recorded spoken message not yet in our audio print database, will be classified as live voice after about 0.3 seconds of speech.
- Turn on auto-answer for agents and use persistent connection with Polycom and WebRTC phones. These actions remove the step in which the agent manually answers the call. See Turn on auto-answer for agents, Find your phone’s configuration settings (for Polycom phones), and Use the persistent connection feature with a Genesys Cloud WebRTC phone.
- Make sure that there are no hardware or headset issues with agents’ phones, slowing connection time. Listen to recordings to troubleshoot these issues.