Starting March 2026, Google Ads converts Lookalike segments in Demand Gen from strict targeting constraints to audience suggestions, reshaping advertiser control over audience boundaries.
Google today announced a fundamental change to how Lookalike segments function within Demand Gen campaigns - a shift that removes the strict boundary-based targeting model advertisers have relied on since the campaign type launched, replacing it with a suggestion-based system powered by Google's AI models. The update takes effect in March 2026 and applies to all advertisers running Demand Gen campaigns in Google Ads.
The change is not subtle. Where Lookalike segments previously enforced a hard ceiling on who could be targeted - capping reach at specific similarity thresholds relative to a seed audience - the updated system treats those same segments as directional signals. Google's AI will use the seed list and selected reach level to prioritize users, but the system may go beyond those parameters if its models predict better conversion performance or lower cost-per-action (CPA).
According to Google's updated Help Center documentation, "Lookalike segments in Demand Gen will evolve from being limited to similarity thresholds to audience suggestions." This represents a departure from the three-tier reach structure - narrow, balanced, and broad - that defined how Lookalike targeting worked until now.
To understand what is changing, it helps to understand the mechanics of the original system. Lookalike segments in Demand Gen were built from first-party seed data, which could include customer match lists, website and app visitors, or YouTube engagement audiences. Advertisers would select one of three reach options, each corresponding to a percentage of users in the targeted geographic location.
The seed list itself had a minimum requirement: the combined total of all submitted seed lists had to include over 100 active matched users. Segments refreshed automatically every one to two days based on available customer data. If the seed list fell below the minimum threshold, the segment could take up to three days to stop targeting - a lag advertisers needed to account for when managing campaign changes.
Segment creation was tied directly to the Demand Gen campaign workflow. Lookalike segments could only be created from within the audience section of an ad group settings page during campaign setup. Advertisers could include up to 10 seed lists per segment, drawing from website visitors, app users, customer lists, Google Analytics audiences, and YouTube audiences. Once created, segments appeared in Audience Manager but could not be edited or deleted from there.
Under the new model, the reach slider - narrow, balanced, broad - no longer functions as a targeting constraint. According to Google's documentation, "instead of restricting your ad group's reach to specific similarity thresholds (2.5%, 5%, or 10% of the target location), the system will use your selected reach level and seed list as signals to help Google AI prioritize conversions."
In practice, this means an advertiser who selected "narrow" to keep targeting tightly scoped to highly similar users may now find the system reaching beyond that threshold when Google's models determine that doing so would improve conversion outcomes. The system will still aim to find similar users based on the seed list, but it will not be constrained by the similarity percentages.
The opt-out option exists, but it requires a form submission. According to Google's Help Center, advertisers who "prefer to keep Lookalikes behavior consistent with the previous Lookalike version" may opt out "by applying through this form." Google states removal takes approximately one week. UI controls to opt out directly from the Google Ads interface are expected later in 2026, but were not available at launch.
A "signal" tag will appear next to Lookalike segments in the Audience Reporting table to indicate when the segment is operating in suggestion mode rather than as a hard constraint.
One of the more technically significant dimensions of this change involves how it interacts with Optimized Targeting, a separate Google Ads feature that expands reach beyond defined audiences when the system identifies likely converters. The two mechanisms now operate in parallel within the same ad group.
According to Google's FAQ documentation, "ad groups with both Lookalike-as-a-suggestion and Optimized Targeting enabled will find higher performance gains being captured by Lookalike-as-a-suggestion mode." The documentation further notes that "performance shift from Optimized Targeting towards Lookalike segment in the Campaign reporting table" should be expected. Optimized Targeting may still capture some gains by reaching users beyond Lookalike-as-a-suggestion's reach.
This layering means that advertisers who have both features enabled are now running two overlapping AI-driven expansion mechanisms simultaneously. The shift toward AI-controlled audience expansion in Google Ads has been a consistent pattern across campaign types in recent years, with Performance Max eliminating manual targeting controls entirely.
Despite the behavioral change, Google has preserved the reporting structure. Performance from Lookalike segments will continue to appear in the Lookalike segment row in Campaign reporting. The signal tag next to that row serves as the indicator that the segment is now operating in suggestion mode. This means historical reporting comparisons should remain structurally consistent, even if the underlying targeting mechanics have shifted.
The minimum seed list size requirement of 100 users for Lookalike segment creation remains unchanged. The process for creating and appending Lookalike segments to Demand Gen campaigns also stays the same. What changes is purely the behavioral logic once those segments are active in an ad group.
Google has also confirmed the change does not affect Lookalike audiences used in other campaign types. According to the company's FAQ, "this change only affects Lookalike segments attached to Demand Gen campaigns in Google Ads. This does not affect Lookalike audiences used on other campaign types like Video brand campaigns or other buying doors like Display & Video 360."
The timing of the announcement - February 2026, with a March 2026 effective date - gives advertisers limited runway to audit their Demand Gen campaigns before the change takes effect. For those who have used narrow Lookalike segments specifically to exclude certain audiences or maintain tight control over who sees their ads, the shift is particularly consequential. The form-based opt-out process is an added step that requires awareness and deliberate action.
The update follows a pattern visible across Google's advertising product roadmap. Google Demand Gen campaigns launched worldwide as a successor to Discovery campaigns, and since then the product has steadily incorporated AI-driven targeting expansion features. The transition from Video Action Campaigns to Demand Gen introduced Lookalike segments as one of the campaign type's differentiating features. Now, less than two years after that introduction, those segments are being redefined in a way that reduces their function as a precision targeting tool.
Adriaan Dekker, a Google Ads consultant, noted on LinkedIn that the change is "another step toward AI-led targeting similar to Meta's direction in recent years," adding that "the request form for hard constraints shows Google knows most won't want full expansion by default." The comment reflects a wider tension in the industry between automation efficiency and advertiser control.
Brian Lasonde, commenting on the same LinkedIn post, described it as "another move toward AI-led targeting," observing that "the request form for hard constraints shows Google knows most won't want full expansion by default." Manoj Kumar, founder and CEO of Rankdriven, took a different view, saying: "Signals over strict limits. AI-led targeting is clearly the future of performance optimization."
Although the targeting logic is changing, Google's guidance on seed list quality remains unchanged - and arguably becomes more important under the new model. According to Google's documentation, advertisers should "use high-intent seed lists" and "ensure your Lookalike segments have users who are the most similar to or closely match your recent converters."
The recommendation to add external converters - users who have engaged with a brand on external media channels - and to combine Lookalike segments with high-intent custom segments that target recent brand or category searchers also remains in place. Under the suggestion model, the quality of the seed list becomes the primary lever advertisers retain for shaping who the AI prioritizes, since the hard similarity percentages no longer apply.
Google recommends creating Demand Gen campaigns and their associated Lookalike segments two to three days in advance of planned ad serving, to allow enough time for the segment to populate before the campaign goes live. That operational detail is unchanged.
The shift mirrors a directional move visible at Meta. Over the past several years, Meta has progressively removed or reduced the emphasis on strict audience boundary controls in favor of broad targeting and Advantage+ audience systems that give the platform's AI more discretion over who sees ads. Google's Lookalike-as-a-suggestion update applies a similar philosophy to Demand Gen.
For the marketing community, this raises practical questions about campaign architecture. Advertisers who built Demand Gen strategies around the assumption that narrow Lookalike segments would act as hard constraints will need to revisit those configurations before March 2026. Those who wish to maintain the original behavior must submit the opt-out form and wait approximately one week for the setting to take effect.
The update also has implications for how advertisers measure the contribution of Lookalike segments. Because performance will continue to be reported in the Lookalike segment row even when the segment is operating in suggestion mode, the data may appear continuous when the underlying targeting behavior has changed. Advertisers comparing performance before and after March 2026 should account for this structural shift in how they interpret their Demand Gen audience data.
Who: Google Ads, affecting all advertisers running Demand Gen campaigns who use Lookalike segments as audience targeting tools.
What: Google is changing Lookalike segments in Demand Gen campaigns from hard targeting constraints - enforcing strict similarity thresholds of 2.5%, 5%, or 10% of the target location - to audience suggestions. The AI system will use seed lists and the selected reach level as signals, but may expand beyond defined thresholds to optimize for conversions or CPA. Advertisers can opt out via a form; UI controls for opting out are expected later in 2026.
When: The change was announced today, February 17, 2026, with an effective date of March 2026. Form-based opt-out requests result in removal within approximately one week.
Where: The change applies exclusively to Lookalike segments attached to Demand Gen campaigns in Google Ads. It does not affect Video brand campaigns, Display & Video 360 buying, or any other campaign type.
Why: Google states the update is designed to improve conversion performance and campaign goal alignment, arguing that strict similarity thresholds limit reach to highly qualified users that AI-driven targeting could otherwise access. The move reflects a broader platform trend toward automated, signal-based audience expansion over manually defined targeting boundaries.
