Average Service Factor...

[cavagnaro]

Hi guys,
I have a customer that is asking for an average on the Service Factor...I don't see how to get this as SF it self is a Category and not a Mask...any idea? Does this sounds logical?

[smile]

AFAIK, SF measure average value already. if you'll check it's values for short time (e.g. take four 15-minutes intervals) and for long time (e.g. 1 hour) you'll see that SF for 1 hour summarize the average value of four intervals.
in such case i ask customer: please take current statistic and show for me what values you'd like to see

[Kevin S]

Service Factor is more of a percentage than an average.
Other than that, I agree with smile - even if you took 15 minute intervals, if you calculated SF over the course of the day (or shift), you would have your "average" for that day.

[tony]

I'm not sure about adding together time periods to show an [i]average [/i]Service Factor... Maybe some context would help..?

If you take Service Factors in time slices, they should be quoted in those time slices.  How would you determine the Service Factor by adding/dividing the following?

15 mins @ 1000 calls with 50 Agents with a SF of 99.9%
15 mins @ 1230 calls with 53 Agents with a SF of 99.8%
15 mins @1098 calls with 43 Agents with a SF of 98.9%
15 mins @ 997 calls with 45 Agents with a SF of 100.0%

Does that mean;

1 hour @ 4325 (TOTAL) calls with 47.75 Agents (AVG) with a SF of 99.65% (AVG)

...No, it doesn't...

Service Factor has been generally agreed as an Industry Standard, to be applied ONCE during a time period.  Any other permutations and/or calculations based on it are, therefore, not Industry Standard.

Taken from the "horses mouth" (ServiceFactor1) =

Value =  100 X nAnsw(TimeRange) / nAnsw + nAband – nAband(TimeRange2)

Basically, you apply the [b]whole [/b]equation to the [b]whole [/b] time period within your Query to show the absolute ServiceFactor.

I hope this helps - or maybe starts a "healthy debate"...?

TT

[cavagnaro]

Thanks guys for your inputs and supporting my initial thoughts. I also challenge the customer on the why he needs that as small samples are the best idea but some new guys that came into the customer wants to "revolution" the stats model ( ) and had some good ideas but also another and mostly weird requests...one of them this average of service factor, according to them to know how actually the SF is...but they already have some historical reports without aggregation and I think that gives them already what they look for...but as it doesn't say AVERAGE implicit they don't agree...
Well will have to have a healthy argue with them and explain as Tony did on numbers they don't lie

[smile]

Tony, ServiceFactor (TotalNumberInTimeRangePercentage) and Servicefactor1 doesn't depends on number of agents in system.

Both of this categories shows the ratio between answered and unanswered (this means calls which was answered outside of SLA range) calls. According to your example:

15 mins @ 1000 calls with 50 Agents with a SF of 99.9% = 1 "conditional unanswered" call
15 mins @ 1230 calls with 53 Agents with a SF of 99.8% = 2 call
15 mins @1098 calls with 43 Agents with a SF of 98.9% = 12 calls
15 mins @ 997 calls with 45 Agents with a SF of 100.0% = 0

that means: for 1 hour we took 4325 calls and 15 of them wasn't answered (doesn't meet SLA)

15/4325*100% = 99,65%

i suppose you can summarize number of calls (and calls in time range) across all intervals and then calculate percentage. Also you can just sum the total percentage across intervals and then just average them.

[tony]

A healthy debate!

As a business, I would not take ServiceFactor as a KPI on it's own - the absolute/overall number of Interactions and the number of Agents would be included in my calculations.  This means I would [i]use [/i]ServiceFactor to calculate my results - but the ServiceFactor itself would not [i]be [/i]my results...

My point is you should not "average" a Service Factor because you will get different results, depending on the time periods involved.  That is the main reason Genesys Volume reporting has Aggregations set to MINUTES, HOURS, DAYS, WEEKS, MONTHS, QUARTERS and ANNUAL - so that you don't have to add unnecessary equations.

In the example then - if the KPI was set very high - at 99.8%...

First 15 MINS = PASS SLA
Second 15 MINS = PASS SLA
Third 15 MINS = FAIL SLA
Fourth 15 MINS = PASS SLA

-but-

First HOUR = PASS SLA

So, really, it depends on the time periods being requested because, if the report is 15-minutes, you would have 75% PASS and 25% FAIL on the KPI - however, if the report was HOURLY, then it was a 100% PASS....

TT

[smile]

Yes, Tony, you're right. Generally speaking average values isn't good for reporting itself. e.g. in hospital: somebody is ill and his temperature is 40 degrees, somebody is died, but the average temp in hospital 36,6 - normal

[tony]

...or worse...

A hospital surgeon writes a report after an operation to remove some cancer.  In his report he says that the operation was a success because the cancer was completely removed - unfortunately, the patient died....

I guess "success" depends on perspective....? 

TT


[quote author=smile link=topic=6414.msg27928#msg27928 date=1305209377]

Yes, Tony, you're right. Generally speaking average values isn't good for reporting itself. e.g. in hospital: somebody is ill and his temperature is 40 degrees, somebody is died, but the average temp in hospital 36,6 - normal
[/quote]