Google Updates Lookalike Audiences in Demand Gen to an AI-Led Model Starting March 2026

Beginning in March 2026, Google will change how Lookalike audiences function in Google Demand Gen campaigns. The platform is moving away from fixed similarity-based audience targeting and adopting an AI-powered suggestion framework designed to improve conversion efficiency and cost performance.

Under the current setup, advertisers upload a Lookalike audience and select a predefined similarity range, such as narrow, balanced, or broad. Campaign delivery is then limited strictly to users who fall within that defined Lookalike pool. With the upcoming update, this constraint will be removed.

Instead, Google will treat the seed audience and selected reach preference as optimization signals, not hard targeting boundaries. The system will be free to reach additional users outside the traditional Lookalike definition if its models predict a higher likelihood of conversion, with optimization focused on conversion volume and target CPA.

What’s Changing in Practice

Lookalike as a Suggestion

Lookalike segments will no longer act as rigid filters. They will function as directional inputs that inform Google’s AI, allowing campaigns to scale beyond exact similarity matches when performance data supports it.

Optimized Targeting Compatibility

This new Lookalike behavior is designed to work in parallel with Optimized Targeting rather than replace it. When both are enabled, the system can unlock an even broader reach by combining audience signals with real-time performance learning.

Legacy Lookalike Audiences

Advertisers who require strict Lookalike-only delivery can still access the legacy setup by submitting a formal request through Google’s support process. This ensures continuity for accounts with compliance or control-driven requirements.

Why Google Is Making This Shift

The change reflects a broader move toward automation-first advertising. Strict Lookalike targeting can limit scale, especially when seed lists are small or noisy. By loosening these constraints, Google’s AI can compensate for imperfect audience data and identify high-intent users more effectively.

This direction closely mirrors recent audience targeting changes by Meta, where advertisers trade some manual control for algorithmic optimization that prioritizes outcomes over predefined audience boundaries.

What Advertisers Should Expect

While this update reduces hands-on control over audience definitions, it is designed to improve overall campaign performance by allowing smarter, more adaptive delivery. For performance-focused advertisers, the shift may result in better conversion rates and more stable CPAs. For those who rely on precise audience constraints, testing and careful monitoring will be essential as the new model rolls out.