METHOD AND SYSTEM FOR ESTIMATING EXPECTED IMPROVEMENT IN A TARGET METRIC FOR A CONTACT CENTER
A system and method are presented for estimating expected improvement in a target metric for a contact center. A lift estimation analysis is performed to estimate the benefit the contact center is likely to achieve assuming different agent availability conditions for a specific future time interval. Historic data is extracted over a set time interval and used to create new datasets for training and testing. The historic data comprises interaction data and associated outcomes for the interaction data. A predictive model is constructed and used to analyze the test dataset by predicting an outcome score for a target metric and estimating an expected lift.
1 . A method of estimating expected improvement in a target metric of a contact center, the method comprising:
extracting a first dataset over a set time interval from a database associated with the contact center, wherein the dataset comprises:
a plurality of past interaction data between customers and agents of the contact center,
associated outcomes for the interaction data, and
future availability of agents from the past interaction data;
creating a second dataset from the first dataset, wherein the second dataset comprises a training dataset and a test dataset;
building, by a predictor training server, a predictive model using the training dataset wherein the predictor training server applies a machine-learning algorithm which seeks to minimize the error in difference between true outcome for a target metric of a given interaction and a predicted outcome of the given interaction;
analyzing, by an analytics server, the test dataset using the predictive model by predicting an outcome score for the target metric for an interaction handled by an agent from a pool of the agents available in the future;
estimating the expected improvement for the target metric when each interaction is handled by an agent meeting a threshold for the outcome score; and
generating, by the analytics server, a visual representation comprising the outcome score for the target metric and the expected improvement.
2 . The method of claim 1 , wherein the associated outcomes for the interaction data comprise customer satisfaction survey results.
3 . The method of claim 1 , wherein the associated outcomes for the interaction data comprise customer satisfaction survey results.
4 . The method of claim 1 , wherein the test dataset comprises a most recent percentage of the first dataset as measured by time over the set time interval and the training dataset comprises a remainder of the first dataset.
5 . The method of claim 4 , wherein the most recent percentage is 20% and the remainder is 80%.
6 . The method of claim 1 , wherein the future availability of agents from the past interaction data comprises a set percentage of a number of the available agents for the contact center.
7 . The method of claim 1 , wherein the machine-learning algorithm comprises decision trees.
8 . The method of claim 1 , wherein the machine-learning algorithm comprises neural networks.
9 . The method of claim 1 , wherein the first dataset further comprises profile information of customers and profile information of agents and the interaction data further comprises a context associated with each interaction.
10 . The method of claim 9 , wherein the analyzing is performed for each of the contexts.
11 . A system of estimating expected improvement in a target metric of a contact center, the system comprising:
a processor; and
a memory coupled to the processor, wherein the memory stores instructions that, when executed by the processor, cause the processor to:
extract a first dataset over a set time interval from a database associated with the contact center, wherein the dataset comprises:
a plurality of past interaction data between customers and agents of the contact center,
associated outcomes for the interaction data, and
future availability of agents from the past interaction data;
create a second dataset from the first dataset, wherein the second dataset comprises a training dataset and a test dataset;
build, by a predictor training server, a predictive model using the training dataset wherein the predictor training server applies a machine-learning algorithm which seeks to minimize the error in difference between true outcome for a target metric of a given interaction and a predicted outcome of the given interaction;
analyze, by an analytics server, the test dataset using the predictive model by predicting an outcome score for the target metric for an interaction handled by an agent from a pool of the agents available in the future;
estimate the expected improvement for the target metric when each interaction is handled by an agent meeting a threshold for the outcome score; and
generate, by the analytics server, a visual representation comprising the outcome score for the target metric and the expected improvement.
12 . The system of claim 11 , wherein the associated outcomes for the interaction data comprise customer satisfaction survey results.
13 . The system of claim 11 , wherein the associated outcomes for the interaction data comprise customer satisfaction survey results.
14 . The system of claim 11 , wherein the test dataset comprises a most recent percentage of the first dataset as measured by time over the set time interval and the training dataset comprises a remainder of the first dataset.
15 . The system of claim 14 , wherein the most recent percentage is 20% and the remainder is 80%.
16 . The system of claim 11 , wherein the future availability of agents from the past interaction data comprises a set percentage of a number of the available agents for the contact center.
17 . The system of claim 11 , wherein the machine-learning algorithm comprises decision trees.
18 . The system of claim 11 , wherein the machine-learning algorithm comprises neural networks.
19 . The system of claim 11 , wherein the first dataset further comprises profile information of customers and profile information of agents and the interaction data further comprises a context associated with each interaction.
20 . The system of claim 19 , wherein the analyzing is performed for each of the contexts.