Service Level Vs ASA vs Abandonment...

[Theo]

Hi all,

Is there a book or website where I can find a correlation between Service Level (SL),ASA, abandonment %, avg wait time, etc. For instance, if the ASA is 20 seconds what should be the SL, abandonment, average wait time, etc. Any feedback is greatly appreciated.

Thanks,
Theo

[JoeB]

Theo,

There is no simple correlation between the variables you have listed. Calls arrive at a contact center in a random fashion (according to Poisson process), some of them enter a queue, and some callers abandon before they receive service. The time to clear each call also varies (typically a long-normal distribution). As a result, the relationships between these metrics are somewhat complex and, in most cases, non-linear, so it cannot be calculated by hand using simple coefficients.

The specific call load pattern (e.g many short calls vs. fewer longer calls), the size of the call center (which impact occupancy rates) and other factors make each situation unique. Moreover, there are human behavior patterns, such as abandonment sensitivity and abandon-retry behavior, that require a case by case calculation of the key performance metrics.

The typical process starts by setting a SL and calculating the other parameters based on call volume, AHT, and the required and available staff. I recommend that ASA is NOT used as a starting point, because even with acceptable ASA individual callers will experience long wait times.

See the article http://www.diagnosticstrategies.com/..._modeling.htm, which explains the theory and the relationships between the various metrics. In addition, you can download software to calculations the numbers in your call center from http://www.diagnosticstrategies.com/easyerlang.htm.

I hope this helps.

Joe

[Martijn]

If you like to know more about the relationship between these variables, see the discussion in the thread: Who can figure this one out.

[thetuan]

JoeB is right in a theoretical sense. But if you have run call centers, you will find that there will be a correlation between ASA and abandonment %. It is true that calls arrive randomly, but you will also find that if you keep good historical data, you will also find that calls arrive at a very predictable pattern. You can get very accurate volume and staffing forecasts on day of week pattern, intraday pattern, and seasonality pattern.

Typically, in regards to ASA and abandonment, I have find that in a customer care environment, if you can get a 20 sec ASA, then your abandonment will be in the 3%-5% range. To get more exact, you have to measure the abandonment rate VS different ASAs measurements within your industry. If you find that more than 80% of your abandonment is occuring after, let's say, 40 seconds and it goes down to 5% after 25 seconds - then you have some empirical data to assume that your customer's abandonment behavior will tolerate an ASA of 20 seconds.

Bottom line is, you need to collect a lot of historical data to start and you need to understand and correctly analyze what you have so you can make proper staffing and scheduling decisions for your call center. If you do this beginning steps of WFM right, you are on your way to become a star!!

Good luck

[JoeB]

Thanks for the comment, thetuan. However, readers may be misled by a statement such as “Typically, … I have find that … if you can get a 20 sec ASA, then your abandonment will be in the 3%-5% range.” Some customer populations may be very happy with ASA of 20 seconds while others may find this wait too long. You may even find that you are being unnecessarily aggressive - I have done work in call centers where ASA of over one minute was quite acceptable. Each user community has different needs and expectations, so other call center ratios may not apply for you.

You need to understand what drives the level of wait time your customers will tolerate (see http://www.diagnosticstrategies.com/abandonment.htm), and, as thetuan suggests, find the abandonment sensitivity of your customers.

This is an opportunity to clarify what we mean when we say that calls arrive randomly. It doesnÂ’t mean that there is no general daily pattern of call volume, but rather that calls do not arrive in an orderly fashion, one after the other, but rather sometime several calls arrive at the same time and while all agents are busy answering other calls, and in other instances no calls arrive for a while. This statistical phenomenon prevents us from coming up with simple rules of thumb to correlate (SL),ASA, abandonment %, avg wait time, etc as requested by Teho.

Joe

[thetuan]

Your comment is correct and that is why the rest of my statement is saying that the call center personel needs to analyze the abandonment pattern to identify the correct ASA for his/her industry. But as I have found out, 20-30 secs ASA is quite tolerable in a customer care environment but anymore than that, then you might have a service level issue with your customers.