Automated Creative Evaluation Based on Multimodal AI for Facebook Reels Ads Using ChatGPT

This work presents a useful multimodal analytical method for automating the creative evaluation of Facebook's Reel ads through the use of ChatGPT. Nowadays, brief ads in digital spaces need to evoke an emotional response from viewers in order to grab their attention. Traditional techniques of evaluating creativity might not be reliable, scalable, or fair. To encourage quantifiable and repeatable innovations, this paper lays forth a method for combining time, space, sound, and motion into a scientifically solid framework. The suggested multimodal analytical framework is utilized. This work explores the themes of time, motion, color, and music. The study utilized notable models through ChatGPT analysis. Video A and Video B are two test ads made with Kling AI and ChatGPT by the researcher. I examined four distinct areas: the visual spectrum, spatial organization, motion segmentation, and the aural spectrum using ChatGPT. I utilized actual Facebook Ad Manager data to demonstrate how the research findings could generate predictions. Video A distinguished itself from the other B with its superior rhythmic movement and overall uniqueness. In contrast, Video B boasted more uniform illumination and superior audio. There were statistically significant differences between the two ads on engagement metrics, including CTR, ETR, and CPC, according to the quantitative data. That was predicted by the researcher through the model results. When evaluating both videos, Video A consistently outperformed Video B. This result provides more evidence that the approach can give us a correct and wise pre-decision before even an A/B test is implied.