Google Demand Gen Now Uses AI for Lookalike Audiences
▼ Summary
– Starting March 2026, Google Demand Gen campaigns will treat Lookalike audiences as AI-driven optimization signals instead of strict targeting constraints.
– This change means Google’s system can expand ad reach beyond the defined Lookalike list to find users it predicts will convert.
– Advertisers can opt out to preserve the old strict targeting model by submitting a dedicated form, otherwise campaigns will default to the new system.
– The update reflects a broader industry shift toward automation, reducing granular manual control in favor of machine-led optimization.
– Google cites reasons for the change, including that strict Lookalike targeting can limit campaign scale and performance.
A significant shift is coming to a fundamental targeting tool within Google’s Demand Gen campaigns. Starting in March 2026, the platform will transform Lookalike audiences from a strict targeting constraint into an AI-powered optimization signal. This pivotal change is designed to broaden campaign reach and enhance performance by granting Google’s automation greater freedom to seek out potential converters, even beyond the predefined audience list.
Previously, advertisers would select a similarity tier, such as narrow, balanced, or broad, and their ads would be shown exclusively to users within that specific Lookalike segment. The system acted as a rigid fence, limiting delivery to that curated pool. Under the new model, these same tiers will function more like a compass. Advertisers provide the initial direction with their Lookalike audience, and Google’s AI uses it as a core signal to explore and target users it predicts are likely to convert, potentially venturing far outside the original boundaries.
This evolution closely mirrors the function of Optimized Targeting, though it does not replace that existing feature. In fact, the two can work in tandem. When advertisers enable both Optimized Targeting and the new Lookalike signals, the system may expand reach even more aggressively. This stacks multiple layers of automation, giving the algorithm increased latitude to pursue campaign goals, whether that’s a lower cost per action or a higher volume of conversions.
For marketers who prefer the traditional approach, an opt-out is available. Advertisers can request continued access to strict, legacy-style Lookalike targeting through a dedicated form. Without submitting this request, all Demand Gen campaigns will automatically default to the new signal-based model when the change takes effect.
This update fundamentally alters the advertiser’s control over audience reach. Lookalike audiences will no longer act as a hard cap on who sees your ads; instead, they will guide an AI-driven expansion. This can significantly impact key metrics like scale, cost efficiency, and overall return on ad spend. It underscores a broader industry movement toward machine-led optimization, a trend already seen on platforms like Meta, where granular manual controls are steadily being exchanged for automated systems designed to maximize performance.
Industry analysis suggests Google is making this move for two primary reasons. First, strict Lookalike targeting can artificially limit campaign scale and hinder performance, especially for conversion-focused objectives. Second, maintaining the complex models required for high-quality, static similarity audiences is becoming increasingly difficult, making a more dynamic, AI-driven approach a more attractive and sustainable solution.
For performance marketers, this represents another step into an automation-centric future. While the reduction in direct control may feel unsettling, similar platform shifts have frequently led to performance improvements for mainstream campaigns. The transition will undoubtedly necessitate a new cycle of testing and learning as advertisers measure the real-world impact on their cost per acquisition, overall reach, and incremental conversion growth.
(Source: Search Engine Land)
