Meta Algorithm Shifts Ad Targeting to Creative
Meta's latest algorithm update, dubbed the "Andromeda Strategy," now primarily uses the ad creative itself to find target audiences, shifting responsibility away from manual media buyer settings. A recent DTC Podcast episode explained that AI now reads creative content to find the right people, requiring broader campaign settings. The update penalizes minor A/B tests as "creative fatigue," favoring distinct, persona-specific concepts over incremental changes.
- The "Andromeda" system is a personalized ad retrieval engine that has shown an 8% improvement in ad quality and a 6% increase in relevant ad recall, shifting the primary performance lever from manual audience targeting to the diversity of creative assets. - Starting December 16, 2025, Meta will use data from user interactions with its AI chatbots for ad and content personalization, creating a new stream of real-time interest signals for advertisers. This will not apply to sensitive topics and will be rolled out in most regions, with Europe being a notable exception for now. - The algorithm now evaluates creative holistically, analyzing imagery, copy, and context in real-time. It penalizes minor creative tweaks by grouping visually similar ads, which can negatively impact performance if one version does poorly. This makes distinct creative concepts—testing different angles, formats, and messages—more crucial than ever. - Meta's "Advantage+ creative" suite automates the creation of multiple ad versions, adjusting aspect ratios, applying filters, and generating text variations to optimize for different placements and audiences. This allows designers to focus on core concepts while the AI handles granular adjustments. - The system's reliance on creative to find audiences favors emerging design aesthetics that stand out. This includes maximalism, bold typography, and authentic, organic-style content that feels native to the feed rather than like a polished, corporate ad. - For freelance designers, this shift transforms the service offering from creating a few perfect ads to developing a "creative portfolio" of 15-20 distinct concepts per campaign that the algorithm can test and scale. This positions the designer as a strategic partner who understands how to "teach" the AI through a variety of creative inputs. - The update reinforces a move away from hyper-segmented campaign structures toward simplified setups with broad targeting, where one campaign might contain numerous creative variations designed to appeal to different user journey stages.
