Zwibbler.com offers a drawing solution for websites. The author criticizes A/B testing as flawed, comparing it to drug trials. Instead, the author proposes the epsilon-greedy method, which adapts to user behavior to optimize choices like button color. This method allows for continuous improvement and automation. The author explains the working principle and benefits, highlighting its potential over traditional A/B testing. Despite the effectiveness of this method, mainstream tools do not support it due to a lack of understanding and education on statistics and machine learning. The author encourages using the epsilon-greedy method for better results.
https://stevehanov.ca/blog/index.php?id=132