An ecommerce retailer was interested in distinguishing which events and actions on their website would most likely result in a sale. In order to better allocate and prioritize efforts for their website, our client asked our data science team to take a deeper look into what steps were most likely to lead to conversions.
Our team quickly recognized there was a marginal impact made by single actions and events, while particular combinations of events led to increased sales.
In order to better understand these series of events, our data science team segmented the web audience into four groups – new and repeat customers, on both desktop and mobile devices. This allowed us to identify which steps were most important for each type of customer in the buyer’s journey.
We fed all of the events and actions through machine learning algorithms to generate probabilities of sales. This consisted of performed analysis on 13,000+ combinations of factors to find the most probable combinations to lead to a sale.
Post-analysis, we learned that content and interaction related events and categories were most predictive of sales. We were able to uncover specific combinations of variables with probabilities of sale as high as 49%.
This information led to immediate impacts for our client. When it surfaced that product site search was consistently a top predictive factor, their team approved additional resources to manage and improve this asset immediately. Due to the high value proposition shown in our research, our client also allocated more time and resources to blog content resulting in even more sales.
Thank you for getting in touch!
Keep an eye on your inbox - We'll be reaching out soon.