Over what period does OCS measure dropped calls?

[Steve]

We are running a new outbound solution, and can see in the logs that the dropped call threshold of 3% is constantly being breached. So we are in and out of predictive like a fidlers elbow.

CC-Pulse reports the dropped rate for the day at 1%, although in a sliding hour it did show us going to 3.2% for a short period.

So I think that OCS does its count over a very short period, does anyone know what it is?

[cavagnaro]

A dropped call is the one that Genesys cuts off, so they are those that exceed the wait time you define in your strategy on the target object.

[Fra]

Cav,
Steve is asking the lenght of time over which the percentage is calculated, not "how" . Interesting question, I'll dig out.

Fra

[cavagnaro]

ups, ok so what and how OCS considers for its dropped level...interesting one...
There is no much info neither on Genesys support site...a ticket?

[Steve]

Cav, yes a ticket might be the answer. But I thought I'd ask here first.

I did go through the Genesys solution search, but as you say there isn't anything relevant.

[Dionysis]

OCS tracks these stats on it's own, independently of Stat Server, so it is not unusual for them to be different.  Also the "overdial" rate that OCS limits dialing on is not only dropped calls, it includes abandoned calls aswell.

Each statistic is held for the duration of the Campaign session.  eg. from the time you load the campaign until the time you unload it. 

[Fra]

Dionysis,

Steve mentioned that the 3% rate is displayed in CCPulse; the stats are sent to the CommDN and then are displayed on CCPulse, so we are not talking about stats calculated in two different components. Question is, what's the period over which they are calculated. My guess is that this is asynchronous.

Fra

[Steve]

It seems that OCS maintains an internal buffer of 120 calls.

OCS stores statistical data in this buffer, replacing the oldest value with the newest values. A prediction using this data is made each second for the number of calls that should be initiated in the next second. It also compares this buffer to the predictive_max_overdial_rate setting presumably in the same 1 second period.

So not only does it measure over a very short period it also uses a very small sample of call data.