Kill it for eCommerce Companies Using Predictive Lifetime Value
Published: March 11, 2014
Author: Ryan Pitylak
Brands that use predictive lifetime value are able to find and acquire more profitable customers.
Facebook’s evolution into a mature advertising platform that helps brands truly meet their business objectives has led to the development of powerful tools for determining return on ad spend (ROAS).
While Facebook allows us to determine revenue we’ve generated for our clients using data that is passed back with a conversion pixel, the capabilities of the site’s internal ad management system end there.
For example, when a user makes a purchase from an eCommerce site, Facebook will only track the revenue that customer generates for 30 days. What about that customer’s lifetime value?
With the rise of sophisticated ad management tools like Nanigans, we can actually find the predicted lifetime value of a particular ad placement. The tools are becoming so sophisticated, in fact, that we can effectively bid on traffic based on the revenue potential of an ad. This drives value for our clients at unprecedented levels.
During the incredibly competitive holiday shopping season, one retailer used Nanigans’ predictive lifetime value algorithms with Facebook’s custom audience product to achieve over 100% same-day ROI and 1,000% 7-day ROI for several days in a row. By bidding on expected lifetime value, they were able to reach the most valuable users during a time of dramatically increased market costs.
The Nanigans platform features several tools and machine learning algorithms to accomplish results like these. When a customer makes a purchase, they are tracked by pixel measurement technology. From this second on, we can use maturity curves to model how this customer will behave over time.
The cohort analysis tool allows us to identify trends over time, while affinity modeling helps us find audiences that share characteristics with our clients’ best customers. Finally, we use the Nanigans layered data and machine learning algorithms to learn which factors drive the most LTV for our clients, so we can optimize ad spend.
For one of our mobile app clients, we used the predictive lifetime value model to calculate the value of our user segments over 45 days. By anticipating which target segments would drive the most return, we gained insight into our most valuable customers. We then targeted these high-value segments, which led to a ROAS above 2.5 and a CPI of less than $1.
By harnessing this information, we can take the concept of providing ROAS for eCommerce marketing dollars to the next level. Because we know sooner and more accurately what the true LTV of an ad will be, we can bid faster, smarter, and earlier.