This scenario describes the simulation of call centers. It is of paramount importance that the correct data is input into the planned WITNESS model so that the expected performance and capability of new proposals for changing building area allocations and staff supervision can be calculated accurately. This analysis shows that these are both key items that affect the performance of the call center teams and therefore for the proposed WITNESS model the data entered for call handling must reflect this.
In this scenario, a company runs a call center that supports their range of products. The center runs 24 hours a day, 365 days a year. On average, there are 300 telephone operators working in the call center at any one time and they are divided into groups to handle different types of call. There are 4 call types:
Pre-sales: product information and advice.
Sales: placing orders and order tracking.
Post Sales / Support: general customer service product enquiries and product support.
Accounts: account management, payment terms, credits/refunds.
Each time a call comes in to the center, the caller is placed in a queue to wait for an operator. The call is logged through the system from this point forwards. When the call is answered, the operator deals with the enquiry; the outcome is decided upon according to the following flow chart (the same process can be applied to follow-up calls and callbacks):
As with all call centers, the response time when answering calls is a key measure of performance. Data has been gathered for a lot of different shifts when there were differing levels of staffing but roughly the same number of calls. A lot of other data was collected too, and appropriate data is selected for analyis as shown in this example.
In a data mining exercise, typical questions might be: what leads to the greatest delays in answering calls? Are there any patterns in the data that could describe the circumstances that lead to delays in answering calls? What causes teams to underperform?
It is important to remember that there will be dominant patterns here - it will be fairly obvious that if you allocate more staff and supervisors to a team, the calls are likely to be answered more quickly. However, the techniques used will demonstrate that you can establish other important trends as well.