The data journalists over at FiveThirtyEight recently posted a controversial article entitled The Hidden Value of the NBA Steal whose central thesis – that NBA players good at something other than scoring can be as valuable to their teams as high-scorers – is a great analogy for explaining the value of multivariate audience data modelling using first and third party data.
What 1st & 3rd Party Data Have to do with Basketball
Sports like basketball are won by scoring more points than the other team, which is why fans get excited about players who score a lot. So what are all those other players for? Teams must win more games when they include some defensive players than when they don’t, otherwise all championships would be won by teams with only offensive players. Basketball is a team sport, and success depends on the continuous interaction of the five players on the court. New analytic approaches to basketball (some of which are mentioned in the FiveThirtyEight article) have exploded in recent years as observers seek out measures and models that shed light on player interaction and individual players' impact on overall success.
If the audience targeted by your performance campaign is a basketball team, then the high-scoring offensive players are first-party data segments: cookie pools rich in converters or interested lower-funnel users. Campaigns that target only this type of segment are known as retargeting campaigns, and they do perform quite well. That said, their value does plateau due to limited reach.
In order for your team to win more games and your campaign to deliver more value, you need to bring in some defensive players: third-party behavioural and demographic data segments that aren’t themselves rich in converters.
These data players complement the more visible stars and your team has the best chance of success when all your players are working together. Whether you call it lookalike modelling, reach extension or segment enrichment, when third-party data teams up with first-party data, your audience grows to include users likely to convert that have never been to your site, while excluding those who have but don’t look or behave like converters. This scales up the performance of retargeting campaigns, just like defensive players in basketball boost the performance of offensive players by stealing the ball and creating scoring opportunities.
Creating a Team to Win, or Target Audience Designed to Convert
Extracting value from third-party data isn’t as simple as just adding some targeting variables to a retargeting campaign, any more than making a championship-winning basketball team is a matter of hiring some good offensive and defensive players and handing them a ball. The players have to be carefully selected to work well together, and with the help of a coach, they learn and practice strategies to win. Machine learning is an audience data model’s coach: machine learning algorithms figure out which combinations of segments work in what circumstances and can help extract the maximum value from each one. And just like a coach helps general managers figure out which players are worth paying to keep, machine learning algos can figure out which segments are or aren’t adding value, helping you figure out which data to invest in.
Want to Know More About Audience Modeling and Targeting?
Learn more about targeting highly qualified audiences with a high likelihood of converting, using the Datacratic Behavioral Data Modeling API.