AI Case Study
Mars Inc improves ad performance with computer vision and machine learning
Mars Inc has leveraged RealEyes's computer vision and machine learning technology to analyse the effect of the choice of soundtrack for an ad. RealEye's system was able to visualise the impact on the audience for two test ads by identifying the emotional reactions of 600 people. Data shown that one track was more effective, giving Mars actionable insight to improve their ad performance.
Consumer Goods And Services
Food Beverage And Drugs
"It took less than 36 hours to test both videos on two sets of 300 people, with the emotional results instantly visible online as they came in.
Track 1 displayed higher happiness levels throughout the ad, demonstrating that it was better aligned with the underlying story than Track 2. Even though the video asset was exactly the same, Track 2 showed significantly reduced average happiness levels.
Track 1 garnered an EmotionAll® score of 5 (better than 36% of ads), much stronger than the 3 (better than 13% of ads) awarded to Track 2, indicating the former will be stronger in driving brand lift metrics.
Mars were immediately able to discern which ad was more successful. When comparing happiness levels between the two versions, Track 1 was unequivocally the more popular and effective version. Using the track would be more likely to drive brand favourability and purchase intent.
"Picking the right soundtrack improved the ad performance x3"
"Mars Inc. wanted to know how the choice of soundtrack to an ad might affect its impact on the audience – so we ran a small study with an old Twix ad to find out.
We tested the same 30-second Twix ‘Conveyor Belt’ ad, but with two alternative soundtracks. The soundtrack choice was particularly important here, as there was a deliberate pause in the middle of the ad to dramatise the moment when the protagonist bites into the second Twix bar. This notable instance made it an especially good ad to test.