Abstract

In today's data-driven business landscape, companies are increasingly leveraging data analytics and machine learning to optimize their lead management processes and enhance customer acquisition strategies. This research focuses on the application of data analytics and the CRISP-DM methodology to improve lead quality prediction and streamline the sales funnel for VAM Consulting, a company specializing in marketing and business technology solutions. VAM Consulting has been developing customer acquisition processes and digital marketing strategies since 2016, amassing a diverse portfolio of over 300 client projects. Despite their extensive experience and utilization of various technologies, the company faces challenges in optimizing lead generation and conversion processes, identifying key characteristics of high-quality leads, and allocating resources effectively to maximize sales efficiency. To address these challenges, this research employs the CRISP-DM methodology to develop and compare three machine learning models - Logistic Regression, Decision Trees, and Random Forests - for lead quality prediction. The project follows a structured approach, encompassing phases of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The dataset, exported from VAM Consulting's Bitrix24 CRM, includes attributes such as lead creation date, stage, source, product of interest, preferred language, and responsible person. Through data cleaning, feature engineering, and exploratory data analysis, the research identifies significant predictors of lead quality, such as lead source, preferred language, and temporal patterns. The machine learning models are trained and evaluated using the preprocessed data, with the Random Forest model demonstrating superior performance. The Random Forest model achieves an accuracy of 0.7372, indicating its ability to correctly predict lead quality for approximately 73.72% of the instances in the testing set. The model also exhibits high sensitivity (0.7444) in identifying good leads and maintains a balanced specificity (0.7103) in identifying junk leads. The research provides actionable insights and recommendations for VAM Consulting, including focusing on leads from recent months, allocating resources to specific source categories (Direct and Marketing), and tailoring communication strategies to cater to leads with a preference for the Arabic language. The study also emphasizes the importance of continuous monitoring, model updating, and compliance with data protection laws and ethical guidelines. This research contributes to the growing body of knowledge on the application of data analytics and machine learning in marketing and sales. It demonstrates how the CRISP-DM methodology can be effectively utilized to develop and deploy predictive models for lead quality assessment, ultimately enhancing business performance and supporting data-driven decision-making. The findings and recommendations serve as a valuable resource for companies seeking to leverage data-driven insights to optimize their lead management processes and customer acquisition strategies.

Library of Congress Subject Headings

Relationship marketing; Target marketing; Sales--Data processing; Machine learning; Logistic regression analysis; Decision trees

Publication Date

5-2024

Document Type

Thesis

Student Type

Graduate

Degree Name

Professional Studies (MS)

Department, Program, or Center

Graduate Programs & Research

Advisor

Sanjay Modak

Advisor/Committee Member

Ioannis Karamitsos

Campus

RIT Dubai

Plan Codes

PROFST-MS

Download

Share

COinS