Realeyes Blog

PreView AI Metrics Linked to Performance Such as Sales & Brand Lift

Written by Colin Pye | 04-Nov-2024 13:45:11

Realeyes’ Human Attention measurement has long been validated against outcomes across the funnel, from awareness to sales. We have now also proved that Realeyes’ AI driven Synthetic metrics predict outcomes for advertisers. 

Realeyes attention and emotion metrics, as well as the Attention Potential (composite score) are powered by our AI models, which are trained on real-life audience reactions, currently over 17 million views across 85 countries. Predictions are tailored by market, audience and ad format. 

This predictive scoring enables partners to verify the efficacy of their creative against their KPI data and then to validate and optimize every ad with confidence. 

 

Sales Lift    Purchase Rates    Brand Lift 

   
Global confectionary CPG    Scandinavian clothing brand    Global OTC healthcare advertiser
         
Attention Potential correlates positively with Sales LIft performance    Assets with higher Predicted Attention tie higher to Purchase Rates   Predicted attention correlates with statistically significant Brand lift
Source 
96 video creatives (TV campaigns) with sales lift data attributed 
  Source 
101 creatives split between video (56) and display formats (45).  
  Source 
88 video creatives (YouTube campaigns) ran in SE Asia & UK

Statistic 
Correlation p<5x10-10 
Lift difference: Mann-Whitney U test, p<3.11x10-5 
  Statistic 
Correlation p<0.03 
Lift difference: Mann-Whitney U test, p<0.15 

  Statistic 
Correlation p<0.011 
Lift difference: Mann-Whitney U test, p<0.012

 

Method 
For those that want to dig into our validation methodology, please consult our Validation Guide. Realeyes offer full-service validation exercises and synthetic ad scoring for your own datasets and modelling.  

Correlated Sales Lift 
A global confectionary CPG studied almost 100 TV campaign video creatives with sales lift data attributed. It found Attention Potential correlates positively with sales lift performance. 

Purchase Rates 
A Scandinavian clothing brand studied 101 video and display creatives, finding that assets with higher predicted attention tied to higher purchase rates. 

Correlated Brand lift 
A global OTC healthcare advertiser studied 88 video creatives from YouTube campaigns ran in Southeast Asia and Great Britain. It found predicted attention correlated with statistically significant brand lift.

 

Cannes Lions

Global brands 

Awardees (proxy for high quality ads) break through Attention Potential norm

Source 
140 shortlisted video creatives that took part in Cannes Lions (2013-2019) 

Statistic 
One sample t-test T=15.6, p<3.6x10-21 

 

 

Associating Quality Creative with Attention Potential 
We studied 140 Cannes Lions shortlisted video creatives from 2013-2019, which served as a proxy for high production quality. We found these videos broke through the attention potential norm. 

The Importance of Modelling 
These results demonstrate why we highly recommend modeling with against all sub-scores, not just the Attention Potential composite score that has been specifically fine-tuned for incremental sales predictions.  

This type of study should be combined with a company’s choice of KPI, such as sales or brand lift data, which should contain the same identifiers as in the ad campaign data to ensure accurate attribution of performance scores and attention data.