Pinterest has rolled out Ad Labs, a new program to trial creative and advertising tools including generative AI features. From personalized background generation to Collages and a Performance+ suite, Pinterest aims to revolutionize marketing strategies and brand safety for advertisers on the platform.
Pinterest has introduced Ad Labs, a new program designed to test prototypes of creative and advertising tools, including generative AI features. This initiative currently focuses on the U.S., where Pinterest is experimenting with personalized background generation for Product Pins using a proprietary AI model. This model leverages Pinterest’s visual search and taste graph to create backgrounds that align with a user’s aesthetic preferences, featuring themes such as “Japandi,” “Organic Modern,” or “Minimalism.”
Another addition is Collages, a feature allowing users to create, share, and shop visual content by cutting, remixing, and reassembling items. This feature is being tested with advertisers like Nike, Bumble, and John Lewis, aiming to make content more engaging and shoppable. Pinterest has already utilized Collages with Wayfair for shoppable Showcase ads, resulting in a 28% more efficient cost per click and a 5.4 times higher engagement rate.
Additionally, Pinterest is developing the Pinterest Performance+ suite, which is in closed beta. This suite combines AI and automation to enhance marketing campaign performance and streamline the creation process, reportedly reducing the required inputs by 50%. Early testing showed a 10% improvement in cost per acquisition for Conversion and Catalog sales campaigns.
Pinterest has also partnered with Integral Ad Science (IAS) and is in the process of integrating with DoubleVerify (DV) to offer brand safety and suitability measurement across all monetized markets. Initial tests demonstrated that over 99% of ad impressions on Pinterest were deemed brand safe. These integrations aim to provide advertisers with enhanced visibility and protection for their brand campaigns on the platform.
