Correlating Neuro-Insight Research with Econometrics

Our work with Company X (we chose to keep the name and other details of the case study anonymous in order to protect client confidentiality) was an in-depth investigation into the true predictive power of Neuro-Insight’s metrics.

 

Starting in December 2013, we pre-tested over 20 ads for Company X. At the same time, Company X was doing research of their own using econometric modeling to look at the actual market impact of the various executions. This offered a unique opportunity for comparison between the two methodologies to understand the predictive impact of the Neuro-Insight approach.

 

We set out to correlate brain response data with Company X’s Mixed Marketing Model (MMM) “Direct Effectiveness” measure, rather than with ROI. ROI is inevitably impacted by the cost of media at different times of year, for example, whereas the effectiveness metric is adjusted to take this into account.

 

There were 17 ads for which we had both Neuro-Insight data and Company X’s MMM effectiveness numbers. We extracted the appropriate numbers for the ads and then entered them onto a spreadsheet allowing us to explore how they correlate, individually and in combination.

 

To understand correlations, we had to start with what we already knew: the best correlation would likely involve emotion and memory. We knew from our academic research and previous commercial correlation work that memory encoding — specifically left brain encoding at branding moments — has a high correlation with sales performance. However, for campaigns focused on creating new emotional associations, memory alone doesn’t tell the whole story. We therefore set out to look at emotional metrics in conjunction with memory.

 

In carrying out the correlation work, we looked at a range of Neuro-Insight metrics alone and in combination. Our starting point was to look at left-brain memory encoding at branding moments, focusing on peaks of response. By peaks, we mean moments when the speed of processing in that brain region shoots up and creates a peak in the data. The most impactful peaks, of course, would fall when branding is on screen. Next, we factored in peaks of emotional intensity and added a final layer, the level of lean-in responses (a.k.a. any positive, approach-like response to the content).

 

Our mixed metric equation became:

(2M + E) * A

where M is the number of memory peaks that feature the brand, E is the number of peaks in emotional intensity, and A is the average level of the lean-in, approach response.

 

In the end, we found an 86% correlation between our metrics and MMM showing the true predictive power of the subconscious mind.