While an agent is conducting an outbound call, CCA uses the predictive dialing algorithm to calculate the time at which the agent's current interview will likely be completed. Furthermore, it also tries to predict how many potential respondents being called will actually answer their phone.

When the number of respondents answering calls is low, CCA will start dialing more telephone numbers to increase the number of answered calls it can pass on to each agent; in this situation, CCA will probably be dialing more calls at any time than the  number of available agents.

CCA estimates the average length of agent calls. Of course, there will be cases when an agent's call just lasts longer than average. This may result in CCA having successfully established an outbound call which it cannot immediately pass on because the initially targeted agent is still busy. In this situation, CCA allocates the call to another free agent. If there are no free agents available to service an established call, the call is automatically broken off. Please remember that potential respondents do not appreciate such "nuisance calls", where they answer their phone, hear no voice at the other end, then hear an error tone.

Ultimately, you can only connect as many respondents as you have agents. When the number of calls dialed increases along with the number of respondents actually answering the calls, you rapidly reach a ceiling determined by the number of agents at your disposition, and all other calls are broken off. It is therefore in your interest to configure CCA's predictive dialing settings to reach a good balance between giving agents as many calls as possible while hanging up on the smallest number of potential respondents as possible.

Predictive dialing results

Askiafield offers several ways to get information on the performance of your predictive dialing system. You can assess the performance during data collection, and when data collection has finished.

Monitoring during data collection

There are three ways to monitor predictive dialing performance during data collection:

• The list monitor function (right-click a sample list and select monitor) displays real-time information about call results. In particular, you can monitor the number of nuisance calls and abandoned by dialer results you are getting. When the maximum percentages of these figures are reached, as set in the dialing properties for the task, CCA will automatically decrease the pace of the dialing so that one call per agent is launched at any time, thereby avoiding the risk to create additional nuisance or abandoned calls.

• The agents window in askiasupervisor shows the average and maximum agent waiting time. When an agent has been waiting for the next call for too long, you can right-click the agent and select no calls?; askiasupervisor will try to determine a reason for the long waiting time.

• For Askia survey projects, you can use the quota availability window (in the tasks window, right-click the task and select quota availability). Here, you can see if a particular quota has been met. When a quota line is full, there could be an effect on the dialing speed.

Monitoring after data collection

Two major reports are available to measure the predictive dialing session’s performance:

• The agent interviewer report contains the ‘success rate’ of the dialing session. This rate represents the number of completes per working time. This report also contains the number of completed interviews, the number of agents that worked on the task, and other useful information. To display this report, open the tasks window, right-click the task and select reports, then agent interviewer report.

• The second report is the task results (including predictive). This report contains the call results and the total number of calls made during the predictive session. To display this report, open the tasks window, right-click the task and select reports, then task results (including predictive) - by survey or task results (including predictive) - by survey (v2).

Data collection tips for predictive dialing

To achieve better performance when using predictive dialing, keep in mind the following points:

• A higher number of agents working in the same outbound group can result in better performance: in general, the more, the better.
• Agents should try to keep as little variation in timings as possible from interview to interview, between the dialing trigger and the end of the interview. For example, agents should not remain on the last screen when they made a mistake but should go to pause. Agents should not open the appointment window on Askia Client and close it several times during the same interview. This will make it easier for the dialler to calculate an accurate algorithm for each agent. Voip cati agents should have the “end call” button enabled during ready state in predictive dialing mode, and hear the ringing sound all the time while they are waiting for an interviewee.
• If you are using the dialing option detect answering machines (found in the dialing properties of the task), then when the “ringing” sound stops, the agent should start talking immediately. This is because the agent will not hear the interviewee picking up the phone and say "hello" if this option is being used.