creative testing
- Past Event Highlights
- Article
Matching Ad Creative Effectiveness and Emotions to Program Content
Peter Daboll – Head of USA, DAIVID
The number of ads continues to explode, but the ability to test them is beyond legacy methods. However, the introduction of AI for ad testing allows all ad creative to be tested rapidly and at a fraction of the price of traditional testing. DAIVID’s Peter Daboll discussed how his company’s AI focuses on human attention, emotions and creative performance to test ads and deliver predictive effectiveness data. Trained with proprietary multi-methodology human input, DAIVID analyzes how people respond to specific features of creative, so future response can be accurately predicted. This methodology includes facial coding, eye tracking and surveys. DAIVID’s measurement approach was demonstrated with a Budweiser Super Bowl ad. DAIVID partnered with the firm Sky to determine the impact of the emotional alignment between programming content and the subsequent ads and found that emotionally congruent placements achieved a significantly higher response rate. Testing lottery and SimpliSafe ads illustrated this finding. Member Only Access
- Past Event Highlights
- Article
Matching Ad Creative Effectiveness and Emotions to Program Content
Peter Daboll – Head of USA, DAIVID
The number of ads continues to explode, but the ability to test them is beyond legacy methods. However, the introduction of AI for ad testing allows all ad creative to be tested rapidly and at a fraction of the price of traditional testing. DAIVID’s Peter Daboll discussed how his company’s AI focuses on human attention, emotions and creative performance to test ads and deliver predictive effectiveness data. Trained with proprietary multi-methodology human input, DAIVID analyzes how people respond to specific features of creative, so future response can be accurately predicted. This methodology includes facial coding, eye tracking and surveys. DAIVID’s measurement approach was demonstrated with a Budweiser Super Bowl ad. DAIVID partnered with the firm Sky to determine the impact of the emotional alignment between programming content and the subsequent ads and found that emotionally congruent placements achieved a significantly higher response rate. Testing lottery and SimpliSafe ads illustrated this finding. Key takeaways:- AI enables advertisers to test every ad in every format.
- Test ads and program content together. Testing together can identify alternative insertion points and compatibility based on emotional signal alignment.
- Maximizing emotional congruence between content and ad drives performance. It also reduces the potential emotional disruption and confusion that interferes with the viewing experience.
- Knowing how a viewer feels emotionally in the moment an ad is served can maximize ad performance while making the viewer’s experience seamless, positive and relevant.
Watch the Presentation
Member Only Access- Past Event Highlights
- Article
The AI Creative Testing Revolution
Duane Varan, Ph.D. – CEO, MediaScience
Duane Varan (MediaScience) explained how adding AI to concept testing allowed more successful ad testing than was possible previously. Testing real video ads against ads created in MediaScience’s Context Studio by their Mediapet.ai platform found that AI ads were on parity in terms of recall, recognition, brand choice, brand attitude, ad liking and professionalism. An Herbal Essence shampoo ad demonstrated the effective use of AI for creative ad evaluation. The original ad was recreated in AI, and creative variables were isolated and tested through the use of “creative twins.” This methodology was validated by Ehrenberg-Bass Institute. Ads for Nespresso, Purina Puppy Chow and Coca-Cola were also successfully tested. Addressable TV can benefit from this approach to target different creatives to relevant viewer segments. Member Only Access
- Past Event Highlights
- Article
The AI Creative Testing Revolution
Duane Varan (MediaScience) explained how adding AI to concept testing allowed more successful ad testing than was possible previously. Testing real video ads against ads created in MediaScience’s Context Studio by their Mediapet.ai platform found that AI ads were on parity in terms of recall, recognition, brand choice, brand attitude, ad liking and professionalism.
Member Only Access
- Research & Data Quality
- Article
2025 TOP MEMBER QUESTIONS with ANSWERS
The ARF Knowledge Center provides secondary research services for ARF and ARF-MSI members on a broad range of topics, especially in advertising, marketing and research best practices. Continuing the core trend from previous years, members tended to ask questions specific to their category or business interests. However, there were also a few hot topics that popped up across membership, such as sports marketing and sponsorships, influencer marketing and AI. In addition, there were some larger, broader trends in themes specific to each constituency, outlined below.
Member Only AccessARF KC Key Takeaways: Ad Testing Methods for Brand Metrics and Creative Effectiveness
Over the past five years, a range of ad testing methodologies to evaluate brand metrics and creative effectiveness have been used by researchers. Approaches include early-stage creative tests like in-context concept ad research, neuroscience-based techniques, including eye-tracking and facial coding to gauge attention and emotion, detailed frame-by-frame content analysis, crowd-sourced feedback and AI-driven predictive modeling of ad performance. These methods help link creative elements to brand outcomes.
Member Only AccessCreative That Converts: A Scientific Approach to Image Effectiveness
Gijs Overgoor of Southern Methodist University introduced a neuro-AI methodology that combines functional MRI (fMRI) neuroscience with large-scale deep learning models to measure and predict how consumers cognitively and neurologically respond to visual content, particularly advertising and product images. The research team first conducted a small fMRI pilot study showing that two neural constructs—cognitive demand (visual processing load) and navigational affordance (brain areas that support spatial mapping)—to explain why some images generate higher click-through or booking intent than others. Because fMRI cannot scale to millions of images, they then built a NeuroVision Transformer, trained on 73,000 fMRI-labeled images, to predict the brain’s region-level responses (≈72,000 outputs) directly from any image at scale, using a neural-network backbone trained on billions of natural images. This enables researchers to produce neurologically-grounded, interpretable features that can feed MMMs, creative optimization systems, e-commerce testing, and product-design workflows.
Member Only Access