In this case study, we will be discussing how we successfully used a multi-armed bandit algorithm for AB testing on landing pages to reduce the cost per lead by 91.63% at a deemed-to-be university in South India. The brand faced a major challenge in terms of the high cost per lead for certain keywords, and needed to find a solution to this problem. By using a combination of landing page tools, an optimized landing page system, and the multi-armed bandit algorithm, we were able to achieve significant results in a short period of time. The following case study will outline the details of our challenge, the solution we implemented, and the successful results that we achieved.
About the Brand
Yenepoya Deemed-to-be University is one of the largest deemed-to-be universities in South India, offering a wide range of academic programs in various fields.
Our Challenge
We were facing a major challenge in terms of the cost per lead for some of our most important keywords, such as “btech admissions in south India” and similar ones. Despite our efforts, the cost per lead was extremely high and we needed to find a solution to this problem.
The Solution
To address this challenge, we decided to use a multi-armed bandit algorithm for AB testing on our landing pages. We used two different landing page tools, Unbounce and another AB testing tool with the multi-armed bandit feature. The latter allowed us to automatically generate and test up to five variants at a time, saving us time and allowing us to conduct a controlled study.
In addition to these tools, we also created our own landing page system using WordPress, which was installed on high-frequency servers from Vultr. We used two layers of caching, Redis cache and Nginx FastCGI, to ensure fast loading times and good performance.
The Result
Within two weeks of implementing this solution, we saw a tremendous increase in the number of action takers on a specific variant. As a result, the cost per lead was reduced by 91.63%, from 800 rupees to 67 rupees, a significant improvement.
Conclusion
The use of a multi-armed bandit algorithm for AB testing on our landing pages proved to be an effective solution for reducing the cost per lead. By automating the testing process and using optimized tools and systems, we were able to achieve significant results in a short period of time.
