New AI Feature for Predicting 'Creative Effectiveness' and 'Target Reach' Implemented in 'AaaS with Meta'

Hakuhodo Inc. (Headquarters: Minato-ku, Tokyo; President: Kenji Nagura) and Hakuhodo Technologies Inc. (Headquarters: Minato-ku, Tokyo; President: Yuzo Tanaka) have launched the third phase of 'AaaS with Meta,' a solution leveraging data and technology from Facebook Japan (Meta), with whom they have a business partnership for brand building on social media. This new feature introduces an AI model that predicts the effectiveness of Meta-specific ad creatives before delivery. By enabling the 'selection of high-performing creatives' and 'pre-prediction of target reach,' it reduces creative verification budgets and allows for optimal budget allocation. Hakuhodo has previously worked to maximize advertising effectiveness in influencer campaigns through the first and second phases of 'AaaS with Meta.' In this third phase, the company is launching a pre-effectiveness prediction function for Meta ad creatives to strengthen creative planning on social media. Meta emphasizes that creative quality is a critical factor in brand image formation and ad effectiveness, and delivering high-quality creatives is essential for maximizing return on ad spend (ROAS). However, traditional operations faced the challenge of not knowing which appeal or material was optimal until after delivery, leading to unavoidable budget consumption during the verification phase. In response, Hakuhodo, winner of the '2025 Agency of the Year,' developed this solution based on its strong partnership with Meta, combining Meta platform delivery data with proprietary AI technology to implement a 'Meta-specialized creative pre-evaluation function' that accurately predicts results before delivery. This new feature uses vast amounts of authorized delivery data to train AI on features such as image size, aspect ratio, and titles. By building dedicated models that reflect the different response tendencies of industries such as automotive, cosmetics, beverages, finance, and food service, it enables high-precision scoring tailored to Meta ads. This allows for the selection of 'high-performing creatives' from multiple candidates before delivery. By focusing investment on these creatives, companies can reduce verification budgets and achieve efficient target reach. Empirical experiments using industry-specific models for cosmetics, automotive, finance, and food service have proven the accuracy and effectiveness of this feature. Initially, delivery models for these four industries have been established. Results showed that delivering only creatives judged as 'high-performing' improved reach costs compared to delivering all materials with the same budget, confirming that pre-delivery predictions align with actual results. Starting with display ads, the company plans to expand to other industries and diverse creative formats. This initiative is also part of the activities of 'HCAI Professionals,' a cross-group AI expert team within the Hakuhodo DY Group. Hakuhodo and Hakuhodo Technologies will continue to combine the AaaS data infrastructure with Meta's technology to accelerate marketing PDCA cycles for advertisers.