service level by interval of day

[jadarvev]

Hello!
I Know service level should be reached consistently through the day, a good KPI for this is to know what percentage of intervals had a service level between an interval(sl+-x). When we calculate this indicator for half-hour intervals and TFS +- 5 points we get that aproximately 30% of the intervals are between (sl - 5, sl+5) (though we achieve de sl diary pretty good.)

Do you know how good is that indicator? or where can I benchmark?

Thanks for your help!!!

[Moe4525]

I would consider using smoothed service level variance as an indices for your service level KPI. This will ensure that during your non-peak hours when Volume or AHT variances are skewed the service level will be smoothed. From the point that the Service level has been smoothed I would then detemine a trend for your contact center. Once this trend has developed you now have a baseline to compare to other centers in your industry. This comparison may be obtained through benchmarking studies from Purdue University online. I hope that I have added some value to this discussion.


Example of "WAPE": {[(Volume *AHT)/1800]/(Sum of 48 intervals workload)}* |(Actual Service Level - Goal Service Level)| = Smoothed Service Level variance

[jadarvev]

Hello! Thanks for ur response. I have some doubts about it.
It´s seems to be an weighted average of the differences between sl achieved and sl goal , where (volume*AHT/1800)/(total workload of day) is the weigt.

Should the formula be aplied to each interval and the Smoothed Service Level Variance of day is the sum of all terms?

Is it O.K the figure 1800 ?, Because with it the sum through all intervals of (volume*AHT/1800)/(total workload of day) is not one.


I appreciate your answer

[Moe4525]

This formula should be applied to each interval. From here you can do a frequency distribution for each interval from a random sample of data points for that particular interval. This will give you a good approximation of your base line for each interval. Of course if your industry has extreme seasonality you will also need to consider that in your random sampling. I look forward to any other concerns or improvements that you may have.

Regards,
Moe