Staff You Need in Inbound Call Center
The dimensioning of an inbound call center involves considering different factors. In doing so, it is very important to identify the goals of care being pursued and determine how many employees will be required to achieve those goals.
To perform this calculation, aspects such as the following should be considered:
- Call forecast
- Average duration of calls
- Hours of operations
- Time distribution of calls
- Attention objectives that can be measured in:
- Level of Care (NA)
- Service Level (NS)
- Average response time (ASA)
- Abandonment (ABA)
- Types of schedules and work shifts.
Workload: The First Step in Sizing a Call Center Inbound
The first step in sizing a call center inbound is to size the number of employees needed based on the workload. To do this, simply multiply the number of calls expected by the Average Operating Time (TMO), also known as Average Handle Time (AHT) for its acronym in English.
AHT is the total average call handling time, which includes:
- The time actually on the phone with a call in progress, which is known as ACD time?
- The waiting time or Hold;
- The administrative time after each call or After Call Work (ACW).
So we have to:
AHT = ACD + HOLD + ACW
In some cases where the telephone exchange is not automatic and the agent must accept the call manually, the RING time must also be included in the calculation. That is, the delay time until the call is accepted.
This measurement must be carried out for each hourly interval, usually measured in strips of 30 minutes or 1 hour. For example, if we expect a forecast of 1200 calls for an interval of 1 hour, with an average operational duration (AHT) of 5 minutes total, we will have a workload or required workload of 6,000 minutes, that is, 100 hours. For this example, we would need a base of 100 agents working during that hour, just to deal with the interval workload or workload.
However, 100 agents will not be enough to meet the objectives of care, since we would not be considering the available time necessary to meet the level of service, nor the reduction of rest times, downtime and absenteeism.
Service Level Calculation: Availability and Performance of an Inbound Call Center
One of the most important aspects in Contact Centers and Call Centers is the measurement of the Service Level. We can define this as an agreement between the client and a service provider. The Service Level (NS) or Service Level (SLA), defines the times and priorities of attention between the different lines of business. This indicator helps to punctuate the levels of availability and performance of the service of attention.
The factors to consider in calculating the Service Level (NS) for the dimensioning of an inbound call center are:
- Number of calls answered, in service level.
- Number of calls received.
The NS is the percentage of calls answered at the service level over the total calls received in a given period of time. The condition for a call to be answered at the service level is its waiting time. For example, if a 20-second goal is defined, those that are attended before said term will be considered as served at the service level, generating the following segmentation:
- Calls answered at service level.
- Calls answered outside the service level.
- Unanswered or abandoned calls.
- Answered in NS + Answered outside NS + Abandoned = Received Calls
- Service Level = (Served in NS / Received Calls) * 100
The NS calculation does not consider the percentage of abandoned calls, so it is usual to combine it with a drop target. The Attention Level (NA) and the Abandonment (ABA) are easier to calculate and are measured as a percentage of the calls answered and abandoned respectively, over the total calls received:
- Attention Level = Attended / Received
- Abandonment = (Received – Attended) / Received
For an objective NS, 2 (two) parameters must be defined: percentage of attention and maximum waiting time. For example, a target NS of 80/20 implies that 80% of the calls received must be answered within 20 seconds of waiting. This is one of the indicators used to assess whether or not the Call Center / Contact Center has the correct number of operators to handle the volume of calls received during the day.
Erlang Distribution: Calculation of the Number of Operators
The Erlang probability distribution helps to calculate the number of operators required to handle a predicted number of calls, in a time interval, meeting a certain level of service and waiting time. It is based on the Erlang C formula (a derivative of the Poisson distribution) that was designed by the Danish mathematician AK Erlang over 100 years ago.
The calculation is very simple, since you just have to specify certain parameters: the number of expected calls, the time span, the average attention time (AHT), the target Service Level (NS), and the maximum waiting time. Once the parameters are defined, the number of agents necessary to meet the target service level can be obtained.
Returning to the previous example, where 1200 calls had to be answered in 1 hour, with an average duration of 5 minutes per call, the Erlang distribution function returns a required 108 agents. That is, 8 more agents compared to the linear calculation of workload.
This difference responds to the waiting time or Avail required between calls and calls to meet the 80% Service Level, preventing more than 20% of the calls received from waiting 20 seconds or more before being answered.
To reinforce the concept, suppose we plan more agents than necessary. We have 200 agents instead of 100. If so, the probability of a call coming in and not finding an available agent tends to zero. The most probable thing then is that no call waits more than 20 seconds before being answered and therefore in service level it will tend to 100%. However, in this case, we will have higher labor costs than necessary.
On the contrary, if we plan fewer agents than necessary; the probability that a call arrives and all the agents are busy will be very high and, as a consequence, the service level will not reach the 80% target, since most of the calls received will wait more than 20 seconds before being answered. In this scenario, in addition to not achieving the objectives of attention and satisfaction levels, billing opportunities will be lost.
In summary, the Erlang distribution allows us to know how many agents will be necessary, above the linear calculation of workload, to maximize the probability of reaching a certain level of service, with the fewest number of agents possible