A core targeting lever in Google Demand Gen campaigns is changing. Starting March 2026, Lookalike audiences will act as optimization signals — not hard constraints — potentially widening reach and leaning more heavily on automation to drive conversions.
What is happening. Per an update to Google’s Help documentation, Lookalike segments in Demand Gen are moving from strict similarity-based targeting to an AI-driven suggestion model.
Before: Advertisers selected a similarity tier (narrow, balanced, broad), and campaigns targeted users strictly within that Lookalike pool.
After: The same tiers act as signals. Google’s system can expand beyond the Lookalike list to reach users it predicts are likely to convert.
Between the lines. This effectively reframes Lookalikes from a fence to a compass. Instead of limiting delivery to a defined cohort, advertisers are feeding intent signals into Google’s automation and allowing it to search for performance outside preset boundaries.
How this interacts with Optimized Targeting. The new Lookalike-as-signal approach resembles Optimized Targeting — but it doesn’t replace it.
When advertisers layer Optimized Targeting on top, Google says the system may expand reach even further.
In practice, this stacks multiple automation signals, increasing the algorithm’s freedom to pursue lower CPA or higher conversion volume.
Opt-out option. Advertisers who want to preserve legacy behavior can request continued access to strict Lookalike targeting through a dedicated opt-out form. Without that request, campaigns will default to the new signal-based model.
Why we care. This update changes how much control advertisers will have over who their ads reach in Google Demand Gen campaigns. Lookalike audiences will no longer strictly limit targeting — they’ll guide AI expansion — which can significantly affect scale, CPA, and overall performance.
It also signals a broader shift toward automation, similar to trends driven by Meta Platforms. Advertisers will need to test carefully, rethink audience strategies, and decide whether to embrace the added reach or opt out to preserve tighter targeting.
Zoom out. The shift mirrors a broader industry trend toward AI-first audience expansion, similar to moves by Meta Platforms over the past few years. Platforms are steadily trading granular manual controls for machine-led optimization.
Why Google is doing this. Digital markerter Dario Zannoni, has two reasons as to why Google is doing this:
Strict Lookalike targeting can cap scale and constrain performance in conversion-focused campaigns.
Maintaining high-quality similarity models is increasingly complex, making broader automation more attractive.
The bottom line. For performance marketers, this is another step toward automation-centric buying. While reduced control may be uncomfortable, comparable platform changes have often produced performance gains in mainstream use cases. Expect a new testing cycle as advertisers measure how expanded Lookalike signals affect CPA, reach, and incremental conversions.
First seen. This update was spotted by Zannoni who shared his thoughts on LinkedIn.
Dig deeper. Use Lookalike segments to grow your audience