PreView gives unprecedented visibility into how attention predicts efficacy of e-commerce ads
A study on creative performance and audience attention
Our client, a multinational electronics company, determines ad campaign success through a proprietary internal conversion rate, a composite measure connecting ad exposure to purchase activity. In 2021, it wanted to maximize the efficacy of the ad campaign by scoring the creatives in advance of allocating media investment.
Our client wanted to increase the ROI of its campaign by identifying the highest performing ads before putting them in market, thereby reducing media waste. It wanted a pre-testing tool that could provide rapid creative diagnostics prior to launch, as well as correlations between attention metrics and the core KPI.
Realeyes used PreView, an AI-powered attention measurement tool, to test both campaign ads and four previous ads from 2020. PreView uses front-facing cameras to analyze facial coding data of opt-in viewers as they watch video content online. PreView produces an ad Quality Score that detects how well an ad captures attention, retains attention, and encodes memory by driving emotional response. It can further differentiate between specific emotional responses such as “engaged,” “happy,” “sad,” and others.
To score the videos, Realeyes convened a panel of approximately 300 people comprising over 1,000 views of four videos from the 2021 campaign and four from 2020.
These tests highlighted strengths and weaknesses in the ads based on attention and emotional response, adding science to creative recommendations. The hypothesis was that the videos’ PreView scores would accurately predict conversion scores.
While the study is preliminary, it predicted the highest and lowest performing ad as determined by the client’s conversion metric, creating a positive correlation between Realeyes emotion metrics and its conversion score.
Given the limited number of ads in this study, the correlation of PreView scores with the client’s metric shows promise and warrant further validation.
Overall engagement to ads was positively correlated (90%), and distraction to ads, its converse, is negatively correlated (-60%) with purchase expectations. In addition, there was an 80% correlation with Encode score.
The highest performing ad from 2021 (according to the client’s internal metrics) recorded the highest PreView retention rate and the highest encoding. It likewise scored the highest in the client’s internal metrics, significantly higher than the second most successful one.
Based on these correlations, we intend to validate further the predictive power of PreView over a larger data set.
Written by Keith O'Brien | Q4 2021