Using Predictive Analytics to Create Business Intelligence Masterpieces
It is hard to believe that both of these paintings were created by the same artist. At left, Picasso’s The Old Fisherman, displays amazing detail and depth of knowledge about the human figure. At right, Three Musicians, presents amazing contrast and is so simple that it almost seems like anyone could have painted it without any training at all, and yet it is the more well-known of the two. I have often wondered why that is.
Believe it or not, there is a strong tie in here with web analytics. The reason that the second image is more powerful is not because of its surface appearance; it is the depth of knowledge of the artist behind it.
Recently one of our largest media clients asked the Adobe consulting team to help them identify four to five key user market segments using predictive analytics and the data captured by SiteCatalyst. This was an extremely challenging task considering the client has over 80 million unique visitors to their site each month, each with unique behavior and content affinities across many individual visits. It was our task to characterize each individual user, and then find a way to group these users into useful and actionable segments that could be used by the client’s marketing team to hone their efforts.
We started by identifying the SiteCatalyst metrics that were most important to our client’s revenue stream. Most of these metrics centered on video measurement but we also incorporated visit information (length of visit, entry time, entry page, etc.), referrer information (including search keywords), and geographic location. We then created an algorithm that could take all available information from each user and create a series of user profiles to which each individual visitor belongs.
Once each visitor had been assigned a usage profile, we used advanced data mining algorithms to crawl through this massive list and identify common usage behaviors across the entire group to create four distinct market segments. Each segment represented unique challenges and opportunities to our client. Some of these challenges and opportunities they already knew about, but they were also surprised at how many opportunities they had not perceived. For example, our client had a very large proportion of users who were coming to their site for a certain video through social references and were not engaging with the site. After looking into this video more closely, our client realized that they were not presenting any additional content to these users. Because of this, they were missing out on a huge opportunity to cross sell additional content. They were also able to identify which user types seem to congregate around certain types of content and thus, significantly educate their marketing efforts.
Many of the recommendations that we made to our client might have seemed retrospectively obvious to a seasoned web analyst. However, like the paintings above, it took an extremely deep dive into our client’s data coupled with an understanding of their market and business model to bring this “masterpiece” to the surface. Even more satisfying to our client was the knowledge that they were no longer relying on heuristics, but were instead creating data-driven strategies around solid analysis.
If you seem to be stuck on the surface of your data and cannot go deeper, contact the Adobe Consulting team for help uncovering the masterpieces below that predictive analytics can bring to light. Speak with your Adobe account or sales representative to learn more.