Hey everyone. I manage our predictive dialer campaigns, and we are getting killed by the FCC abandonment rules right now! Our connection rate is decent, but when a live person picks up the phone, there is a massive 2-3 second delay before the agent actually hears them. The customer gets confused by the dead air and hangs up! I have Answering Machine Detection (AMD) turned on because we do not want agents wasting time on voicemails, but the AMD evaluation is taking way too long. Is there any way to speed up the AMD analysis or bypass the delay while keeping predictive dialing?
Greetings. This latency is an inherent architectural limitation of Answering Machine Detection. The system must sample the initial audio energy (the ‘hello’) and analyze the cadence to determine if it is a human or a recording.
If you are utilizing Genesys Cloud Voice, the media must traverse the AWS infrastructure for this analysis. To significantly reduce this latency, you must transition from a pure cloud telephony model to a ‘BYOC Premises’ architecture using local Genesys Edge appliances.
By executing the AMD DSP (Digital Signal Processing) locally on the Edge hardware rather than routing it through the cloud, you can reduce the evaluation latency to under 400 milliseconds.
I must interject regarding the regulatory implications of your current configuration. Under the strict guidelines of the Telephone Consumer Protection Act (TCPA) and the UK’s Ofcom regulations, an outbound predictive dialer must connect a live individual to a human agent within two seconds of the customer’s greeting. If your AMD evaluation is causing a three-second delay resulting in abandoned calls, you are in direct violation of the statutory limits.
I strongly advise you to completely disable Answering Machine Detection until your underlying infrastructure latency is resolved, to avoid significant financial penalties.