Telecommunications Call Center Labor vs Call Volume Balance
IAS worked on an Telecom engagement where they needed to balance their call center labor force to the call volume coming into the center. IAS had worked on credit risk modeling project for this client previously.
IAS observed the call center volume was driven up by the telecom turning off customers phones. Customers phones were being turned off because they were leaving to go to other telecom competitors or because they did not pay their bills. Customers not paying their bills were assigned a risk score and placed in the que to be turned off.
IAS captured the customers affecting the call center volume who were not paying their bills. We gather a time series dataset of the volume of customers in various states of phone cancelation by their risk score. We constructed a periodogram to understand the cycle of call volume for each of these stages. Then we, used Spectra Analysis to describe the correlation between timeseries most effecting the call volume. We were also able to determine the time from when a customer entered a risk state for cut off and when call center volume would be affected.
This project led to the client being able to better manage their call center labor and was awarded the J.D. Power for best call center in the industry for customer support that year.
A major manufacturer hire IAS for and engagement to report the various sales cycle for their computer hardware products. Many time series show periodic behavior. This periodic behavior can be very complex. Spectral analysis is a technique that allows us to discover underlying periodicities. To perform spectral analysis, we first must transform data from time domain to frequency domain. the covariance of the time series can be represented by a function known as the spectral density. The spectral density can be estimated using on object known as a periodogram, which is the squared correlation between our time series and sine/cosine waves at the different frequencies spanned by the series (Venables & Ripley 2002).
Above is a periodogram showing a 1 year period as the dominate cycle.
I build periodograms for three of their product and reported each products sales cycle.
Coherence is a time-series measure similar to correlation. The next step was to to check if there was an association between products based on their cycle. IAS took the coherence between the products and examined the cross-correlation between them. There was clearly a significant correlation in these products lags.
This engagement left the client with a tool to enhance there forecasting portfolio and allowed them to design new strategy to response to know that this sale cycle behavior was occurring and that there was a relationship between two of their products the illustrated lead lag behavior .