Vedantu Retention: Testing a Human Support Model

During the COVID lockdown, Vedantu saw a rapid increase in online enrollment. Students between 12 and 18 were joining courses while also learning how to navigate the product, and the same landing experience served free and paid users.

I worked with the product team on the paid-user journey and landing page. We tested changes intended to make course navigation and discovery clearer for premium students.

The customer-success question was whether direct human support could reveal and resolve problems that a ticketing system might miss. We created WhatsApp groups for selected JEE batches and placed a customer success agent in some cohorts while leaving others on the existing support model.

The test helped us see the practical work agents were doing for students. They could identify navigation problems, answer questions in context, and surface recurring pain points. Those observations informed the expansion of the customer success organization to more than 200 agents.

Retention moved from roughly 33% to 45% during this period. Several interventions were running at the same time, and lockdown conditions also affected behavior, so I would not attribute the full change to our test.

I also developed a roadmap for moving parts of the interaction model toward WhatsApp-based AI chatbots, including tools such as LimeChat and Yellow.ai. I left for another opportunity before that automation was deployed.