The persistent demand in most businesses is knowing what drives consumer behavior. Two important trends contribute to the burgeoning power of prediction in the market research industry: new data sources and new analytical technologies. ENGINE fits nicely into the nexus of these two trends; like building the “last mile” embedding market insights directly into campaign implementation.
New Data Sources
First, the availability of new data sources is tied to the growing ease with which non-survey data can be appended to any survey. This is linked to the growth in first-party data capture (e.g., CRMs tied to loyalty programs and purchases) and the proliferation of third-party data collection from online and off-line sources which can be connected via digital pixels.
While first-party data remains an analytical keystone, first-party data is inherently limited to the consumers with whom a brand is already doing business. It is the possibility of third-party data integration which extends both the scope (i.e., entire market) and the predictive power of the survey. Some of the more useful third-party data captures extensive digital activity (e.g., site visits), content consumption and location patterns. The behavioral nature of this third-party data is an effective compliment to the motivation and intentions that can be gleaned from survey responses. It is the relationship between beliefs (i.e., intentionality and motivation) and behavior that makes prediction using integrated data so promising.
New Analytic Technologies
The second important trend is the emerging plethora of machine learning technologies, which are more robust and more adaptive to the unique prediction problems in the market research industry than ever before. While the statistical theory and mathematical processes underlying the vast array of available machine learning technologies can be bewildering, the straight forward method of “train and test” means that business decision makers can easily assess the predictive potential of any analysis by simply looking at the misclassification table. In other words, how many times did the algorithm incorrectly or correctly predict a business relevant outcome using the integrated data.
Finally, there is business value in using predictive algorithms which are built in an integrated environment to enhance ROAS (i.e., return on advertising spend). For example, marketers can use prediction algorithms from integrated environments for campaign targeting. The prediction algorithm can be used to build a high-performance audience in the third-party’s data universe. This audience can be ported to a DMP or ad exchange to ensure more effective targeting than traditional Boolean alternatives.
The advantage of using this approach with integrated data is that you get the forward-looking capability of survey-based intentionality, combined with known behavior. The combination of these two types of data provides a customized prediction algorithm that is testable and unique to those that control both sides of the equation: belief and behavior. For the Engine Group this translates into competitive advantage for our clients. This is the promise of augmented ROAS in an increasingly undifferentiated programmatic advertising world.
Learn how ENGINE leverages data in the areas of Market Research and Customer Experience Research.