An Introduction to Predictive Dialling
Predictive dialling is a powerful method that increases the talk time per hour per agent.
Definition
Predictive dialling is a method to dial more calls than the number of agents that are available, in order to bring down agent waiting time to a minimum. The exact number of these calls is determined dynamically depending on the current situation, the previous calls status and the limits set in the user interface of the Supervisor application.
Settings, Definitions and Recommendations
| Setting | Definition | Recommendation |
| No Answer Time | The ringing time before hanging up and dropping the call as a "No answer". | 20 to 30 seconds (3 rings) |
| Max simultaneous calls | The number of channels to reserve for the predictive dialling. |
2 x the number of agents allocated to the Survey task (e.g. 10 agents =>20 simultaneous calls) |
| Max nuisance call level (%) |
When the respondent picks up the call, but the dialler drops the call because there are no agents available. If this % is reached the dialler will dial less aggressively, in order to not go over the max value. | A value of max 1-2% is allowed (but check your national legislations). |
| Nuisance ceiling time (s) | Before dropping the call, the dialler will wait for X seconds to see if an agent will become available. |
For e.g., if the nuisance ceiling time is set to 5s, a contact will be kept on the line for 5s in case the agent becomes available during this period. Please note that generally, a respondent hangs up after 5s if there is no voice to let them know that an agent becomes available . |
| Max abandoned call level (%) | When the dialler abandons the call before the no answer time and without having someone on the line. This means that a ringing phone on the respondent side will stop ringing before someone picks up. The ‘Max abandoned call level’ setting will manage the maximum percentage of abandoned calls allowed. |
The max value is 10x the maximum nuisance call level (%). A value of 10 to 20% is common. If set higher, the performance of the predictive dialling algorithm will be better. |
|
Optimal wrap-up time (s) |
The time between two calls (before passing through another call to the agent) |
Between 0 to 3 seconds |
|
Detect answering machines |
Allow the dialler to detect if the connected call is an answering machine or not. The detection is based on speech intonation and patterns. | However, agents should be trained to start talking right after the ringing tone drops. This is because agents will not hear the “Hello” from the respondent when this option is enabled. |
|
Detect faxes |
Allow the system to detect fax machines. | |
|
Enable predictive triggers |
This will enable the triggers that are set in the questionnaire or script. For more information on predictive triggers see best practice. |
With a predictive trigger, the waiting time will be reduced. If there isn’t a trigger set, the system will automatically use the last screen as a trigger. |
|
Group based predictive |
If the group based predictive setting is enabled, dialling calculations will be made for the entire group working on the same task.
|
We recommend turning this off if there is some variance between time and interview length by agent. CTArchitect will automatically activate agent-based predictive dialling. This mode lets the dialler algorithm keep parameters for each agent individually. |
|
Answer Machine message |
Play an automated message on a detected answering machine: the text field allows you to fill in the path, name & location of the sound file you want to use. | Use .WAV format. |
| Nuisance message |
Play an automated message to a respondent when the nuisance ceiling time is exceeded: the text field allows you to fill in the path, name & location of the sound file you want to use. Interruptible checkbox: when this setting is enabled, the call can still be passed on to an agent when he/she becomes available during the playing of this message. |
Use .WAV format. |