How Call Analysis detections affect Abandon Rate

This topic provides supplemental information about Call Analysis Detections as they affect calculation of the Abandon Rate statistics.  The method used to calculate Abandon Rate can directly affect agent idle time and the speed of outbound dialing.  The Detections option calculates abandon rate in a way that plans for less-than-perfect Call Analysis.  The best way to understand this option is to consider a real-world scenario.

Suppose that 250 numbers are dialed, and that 150 of those calls are auto-dispositioned.  The remaining 100 calls are classified as live people by Call Analysis.   Dialer routes those 100 calls to agents, but 3 of those 100 calls are abandoned because there are no available agents to take the call.  Of those 3 abandoned calls, 2 were actually live people and 1 was really an answering machine.

The other 97 calls are dispositioned by agents.  Of the 97 dispositioned by agents, 78 calls were actually live people and 19 are not live people and are marked as such by agents (17 answering machines, 2 SITs).  Agents almost always correctly identify a live person versus an answering machine, so their "call analysis" is very accurate.

Using the standard mapping of what is a contact and what is not (as configured in Interaction Administrator), the number of contacts is 81 (78 plus the 3 abandoned calls,  since System HangUp is a WrapUp Category).  In this case the abandoned rate as calculated by Dialer is (3 / 81) * 100 = 3.7%.

This is not the actual abandon rate, since Call Analysis is not perfect.  If it were perfect, no answering machines would go to agents or get abandoned.  So, the actual abandon rate is (actual live people abandoned / actual live people detected) or (2 / 78) * 100 = 2.6%.

Poor call analysis at a Dialer site affects abandon rate, since the number of contacts decreases while the possibility of an abandoned call from calls routed stays the same.  This can cause contact rate to decrease and abandon rate to increase, leading to slower dialing and more agent idle time.

To alleviate this situation, use the "Based on Detections" option to make the "best effort" calculation possible with current technology.  This approach considers that the system thought that 100 calls were live people, and 3 of those system-detected live people were abandoned.  It recognizes that the system has no way to know that 1 of those 3 system-detected live people that were abandoned was really an answering machine.  Therefore, due to a computer's less-than-perfect ability to detect live people, the best calculation of abandon rate in Dialer would be:

(system-calculated abandons / system-detected live people) * 100

or

(3 / 100) * 100 = 3%.

The Detections calculation returns a value (3%) that is lower than the abandon rate that would normally be calculated by Dialer (3.7%), and which is closer to a perfectly calculated abandon rate (2.6%) which cannot be determined, due to less-than-perfect ability to detect live people.