Lead Conversion and Scoring with Machine Learning

It has been proposed in this paper a solution supported in the Machine Learning approach. This solution is implemented to increase the winning probability of leads. The Logistics Regression has been selected as a final with accuracy above 80% with the value of area under the ROC curve (AUC) is high (0.89). “Lead Score” >  = 85 are hot leads and should be targeted first. Around 80% of total conversions could be attained by targeting 50% of the total client base. There are 13 important features from the final model. The top 7 features have a positive impact on lead conversion. The bottom 6 features have impacted lead conversion negatively. Deep Learning models like Deep Neural Network (DNN) and Sequence Model (LSTM) can be tested on both original data and additional data for future scope.