I’m currently working as an Area In Charge at National Gas Company and am primarily responsible for generating sales revenue with the primary aim of achieving and increasing profitability as per set targets for projects, commercial propane and bulk LPG in the Oman region. In this data-driven corporate world, companies thrive on data collection to generate insights in order to sustain and grow. Courses which feature Data Science and Business Analytics equip an individual with the necessary skills required to be efficient in data analysis for insights and also help with career transition.
My company was generating sales data on regular basis however, the same was not being utilized to derive actionable or resourceful insights. The data collected was merely used to prepare daily MIS and yearly projections. The data could have been used to identify sales trends, customer churn analysis and even more. My company was facing tremendous pressure from the market to generate revenues due to higher customer churn. A higher rate of customer churn was affecting the cash flows and revenue generation of the company. Due to this, my company was losing its market share and its position as a market leader. I thought of utilising Machine Learning and Predictive Modelling.
In order to address the concern of customer churn, we had to predict beforehand whether a customer would leave us or continue business with us. To predict whether a customer would churn or not we resorted to Machine Learning and Predictive Modelling to build models which would predict whether the customer would churn or not from the available data. Data collection was easy enough but data cleaning and selecting the features which would help me to build the models posed the biggest challenge as employees with data science and business analytics skills were only me and my marketing manager. Data analysis revealed some insights which were shocking and enlightening at the same time. The sample we selected to analyse was imbalanced as it had 90% non-churning customers and only 10% churning customers. I had to apply SMOTE to balance the sample dataset so that models would perform better on the test dataset. Post analysis we were able to see the features which my company needed to focus on to avoid further customers from leaving us.
Product pricing came out to be the game changer for my company, which also turned out to be a major contributing factor towards customer churn. This exercise not only helped us to redesign our product pricing but, in the process, we ended up optimizing our supply chain operations which in turn improved our customer experience. This also ensured safeguarding our position as the market leader while simultaneously maximising our revenues. My company immediately reworked our product pricing by optimizing our supply chain and negotiating thoroughly with our vendors and we saw a change of 12% in customer churn by all having this exercise. This exercise helped us in identifying the potential customers who would churn in well in advance and helped us in devising the countermeasures to retain the customer. The data analysis also helped us to identify the new segments and geographical locations my company can target for future business expansion.
