VAG/SkillName Routing and Overflows... How to report on Target Delivery?

[tony]

Hi!

I'm looking at a rather complex routing strategy, which works generally in the following way, for the Delivery of a call;

Background:
Agents are attributed with a SkillName and SkillLevel, which automatically applies them to a VAG with a naming convention which reflects the same SkillName and SkillLevel

Delivery Routing:
1. Primary Target: Attempt to Route to Agents in VAG_SkillName_SKillLevel[b]1[/b] for 10 secs.
2. Secondary Target: Attempt to Route to Agents in VAG_SkillName_SkillLevel[b]1[/b] and VAG_SkillName_SkillLevel[b]2[/b] for another 10 seconds.
3. Backup Target: Attempt to Route to Agents in VAG_SkillName_SkillLevel[b]1[/b] and VAG_SkillName_SkillLevel[b]2 [/b] and VAG_SkillName_SkillLevel[b]3[/b], indefinately.

This works well, from a Routing perspective.  Unfortunately, there is a problem when trying to report the SkillName and SkillLevel used for the Delivery of the call, in CCA;

If the call is Delivered under scenario 1., the reporting works fine.
If the call is Delivered under scenario 2. or 3., it depends on which Agent took the call, as to whether the reporting is accurate.  For example, here are 2 possibilities for scenario 2;

A. The call waited 12 seconds and an Agent with SkillName_SkillLevel[b]2 [/b] becomes available and takes the call.  Call is Reported via VAG_SkillName_SkillLevel[b]2[/b] - which is good, since it was delivered to the Secondary Target.
B. The call waited 15 seconds and an Agent with SkillName_SkillLevel[b]1 [/b] becomes available and takes the call.  The call is Reported via VAG_SkillName_SkillLevel[b]2[/b] - which is not good, since it was delivered to the Primary Target, not the Secondary Target.

I expect that this is an age-old issue and the "resolution" is to report the correct Skills via CCON/ICON. So - if that is [i]"not an option"[/i], how would we achieve accurate VAG reporting, in CCA?

Thanks!

Tony

[tschoi]

Hi Tony,

I don't know exactly on which metric you're reporting on but the way to get what you want is following:

1. VAG_SkillName_SkillLevel1 should have skill expression of skill=1
2. VAG_SkillName_SkillLevel2 is skill=2, so on and so forth.

When you design a strategy, you need three target block. The first one will target VAG_SkillName_SkillLevel1 and have timeout of 10 secs. There is a checkbox called "Clear Target". Clear this checkbox. The second one targets VAG_SkillName_SkillLevel2 and have timeout of 10 secs. Also clear the "clear target" checkbox. The final one has VAG_SkillName_SkillLevel3 and very long timeout (few hours?) cleartarget does not matter here.

In this kind of strategy, URS considers SkillLevel1 and SkillLevel2 as possible targets when it's on second target block. But it there are two advantages.

1. If you specify different VQ for three target blocks, URS will clearly mark which VQ was used (which target was selected) and StatServer can count it.

2. You can easily count calls for each VAG as they don't overlap.

Cheers,
Taisuk

[tony]

Makes absolute sense... thank you very much!

...drawn a little picture, just so's its clear...

Tony

[cavagnaro]

Remember only to create the StatType Cleared.

[tony]

I think I get it - make sure the StatType is cleared, even though the Target is not, to ensure that the VQ stats reflect the VAG the call was routed under...

?

[cavagnaro]

Sorry if not was clear, I ment to do this on StatServer:

Create a metric with MainMask:
CallsCleared—number of calls diverted from one virtual queue to another.

Using One Virtual Queue per Skill
If one virtual queue is assigned to each skill, then skill-specific queuing statistics can be generated.
If you select Clear Target during target configuration, the call leaves the virtual queue completely when the target (a group of agents who happen to have the specified skill) associated with the virtual queue has a timeout parameter that expires. In this case, agents with the specified skill no longer can answer the call if they become available after the timeout period.
Note: Genesys recommends that you do not select the Clear Target checkbox during target configuration if you are using skill-based
reporting with one skill per virtual queue.

Using One Virtual Queue per Skill Level

Another approach to skill-based reporting is to use different virtual queues for each target list in the routing strategy. For example, a strategy could specify that the first target list includes billing agents with a high skill level, the target list following a timeout  includes medium level agents, and a third list includes all billing agents. By using a different virtual queue for each target list, the
client can measure how many calls waited in each queue and how long they waited. However, this type of configuration can make the interpretation of the data quite difficult in CCPulse+. The remainder of this chapter attempts to clarify the issues with a series of hypothetical calls and the resulting statistics for eachof three virtual queues, BillingHigh, BillingMed, and BillingAll. All scenarios
assume that a distributed call rings for 3 seconds before being answered. Also, whether you select or clear the Clear Target  checkbox in the Target Selection dialog box in Interaction Routing Designer significantly affects show Stat Server measures calls. The next two sections compare the reporting of various calls with this setting both selected and cleared.

There is a very good example at statserver reporting reference guide on Reporting on Virtual Queues section.