Where most people shriek away from ‘difficult Erlang C’-methods to determine the needed staff for customer service, find below a way to do it on the back of a napkin.

Background

The creation of a budget follows several necessary steps in a particular order, which I call the ‘budget chain’. Please find below a depiction of how one starts with the Average Handling Time to combine it with the number of transactions to get to the Total Handling Time (or Productive Time), and so on up to a complete budget:

AHT x #transactions => Productive Time

+ Available + shrinkage => logtime

+ overhead hours = >total paid hours

x average salary level => total labor cost

+ overhead charges => total cost

(+ profit margin = >sales price)

As a famous Dutch soccer player once said: “You will only see it when you understand it”, and I expect that the chain above might help several to understand the process. Most experienced Customer Service commercials have certain percentages in their head of how to get from one to the other, and in my previous blog I covered the relationship between the Total Handling Time and the Paid Hours. Below I will show an easy way of calculating’what some consider to be the most difficult component in that relationship: the determination of the Ávailable Time.

Determination

The very first question of a manager when confronted with a new service should be about what kind of transaction the agents need to perform, the kind of background and training is necessary, and the average length of such a transaction. The qualitative elements focus on the (background and education of) people to recruit, which I will leave aside here, the quantitative one about the Average Handling Time is the only one relevant here.

The second question one might expect is about how many transaction per time period we are talking. With these 2 components we can determine the Total Handling Time that is required to provide the service. So far so good.

Together with the answer to the third question about the required Service Level, one can now calculate the required workforce. And this is exactly where most people in the business I have met ‘log off’. Many are afraid of extremely intelligent calculations based on incomprehensible Erlang C formula’s. To what extent these predispositions are fed by providers of services in this area I do not know, but these assumptions are definitely unfounded.

Front Office example

Let’s take an example of a standard customer service for Product A, where customer contact the organization for general questions, which require an Average Handting Time of 180 seconds, so 3 minutes. I presume none of my readers is so ‘out of it’ that she will conclude that therefore the agent can handle 20 calls per hour, so the next step becomes the determination of the number of agents required, given the Service Level desired and the volume of calls per hour. Let’s say that the Service Level is determined to be 80/20, which means that 80% of the calls need to be answered within 20 seconds. This Service Level – conbined with the volume of calls – determines the ‘inefficiencies’ of the waiting queue, so the required ‘available time’.

For the calculations below I used the Erlang C-calculator from Prof. Dr. Ger Koole of the Amsterdam Free University (VU) at http://gerkoole.com/CCO/. This calculator nicely shows how many people are required in relationship to the volume of calls per hour.

Note: AHT= Average Handling Time

THT=Total Handling Time

Occupation = THT/(THT+Total Available Time)

Shrinkage = 0 (unrealistic but to keep the calculation simple)

The crucial number is the one in the ‘Occupation’ column, which indicates which percentage of the used hours is really applied to Productive Time. For ease of use I sidestep for now several complications, like e.g. the discussion about shrinkage, the fact that volumes are not equally divided over the day/week/month/year, and the efficiencies if agents can answer multiple workgroups, e.g. like e-mails. Also, please note about which hours we are talking here. (“apples, pears or oranges…”)

In short, one can take the Total Handling Time required, divide it by the estimated THT/log hours from the table above, to get to the log hours required. Using the numbers for the previous blog “apples, pears and oranges…” one would then take 15% shrinkage, so the log hours in the table are multiplied by 1/1.15.

Depending upon once overhead assumptions, the budget then follows handily.

Back Office

Often, it is supposed that a non-call department would not have any available time at all. Unfortunately, that is usually not the case. Below an example using the same calculator for a Service Level of 90% answered within 12 hours, with a 10 minute AHT. This example shows that the need for Ávailable Time in e-mail is very much dependent upon the variables like (Service Level and) volume. In case a supplier does not take into account the 17% Available for that small workgroup and this does not get funded, we here have the basis for an ever growing backlog which will lead to more follow-up telephone calls and e-mails and thus a devil’s circle of ever growing work.

Back Office activities have some peculiar challenges of their own: very often breaks are not registered properly (“Oh, sorry, I forgot to log”) and a functional check with a colleague often has an ‘interpersonal component’ there as well. It is not uncommon for agents to prefer to work on e-mail, after which they become unmanageable to go back to the phone. Therefore, it requires a special skill for a manager to deal with these challenges in the long run. Mathematically, though, it goes exactly the same as with the Front Office.

Interestingly enough, social media typically require very fast response times. This means that the Service Levels of these services closely resemble the ones on calls. Combined with the very low number of volumes that are common these years make the Occupation levels so low that these jobs need to be staffed similarly to ‘standard’ office workers. As such, these jobs then provide great growth paths to other administrative kinds of jobs.

Next blog, I will expand on the Overhead Charges, where many a customer service organization goes awry in my opinion

(Originally published September 12, 2013)