Enhancing Ad Testing Via New Techniques

Summary

Single-source data allows analysts to match what consumers watch with what they buy.  Using Analysis of Covariance models (ANCOVA) enables researchers to measure the impact of advertising on store sales in test vs. control situations.  But sometimes controlling for the ad exposure and past behavior variables in an ANCOVA can get messy. Nielsen Catalina Solutions (NCS) developed Cognitive Analytic tools to help control for these factors and other targeting biases.  NCS has trained a series of models to choose the optimal analytic approach applying a combination of these variables:

• Purchase Behavior
• Exposure
• Timing of Visit
• Brand Purchase
• Media Consumption
• Category Purchase
• Location
• Demographics

NCS applied these techniques to over 3,000 cases in their database.  When these Cognitive Analytic tools were compared to traditional test vs. control methodology, they found that:

• No systematic variations results; differences are distributed around 0
• 61% of results had a difference of $0.01 on incremental dollars return per exposed HH basis
• A high degree (78%) of intersection in confidence intervals was found

Where differences existed, they could be explained by the additional rigor of Cognitive Analytic techniques.