How to Create a Lookalike Audience in Google Ads?

Can you create a “lookalike audience” in Google Ads?

If you’re coming from Meta (Facebook/Instagram) or other paid social platforms, it’s natural to look for a button that says “Lookalike.” In Google Ads, there isn’t a like-for-like “Lookalike Audience” feature you manually build and then target forever.

Historically, Google Ads offered “Similar audiences / similar segments,” but those stopped being generated for targeting and reporting starting May 1, 2023. In practice, the modern Google Ads equivalent of lookalikes is achieved through automated audience expansion using your first-party audiences as the starting point (your “seed”).

The two modern “lookalike-style” options you should use instead

Optimized targeting is the workhorse for performance-focused prospecting on Display, Demand Gen, and certain Video campaigns. You provide signals (like Customer Match lists, remarketing segments, custom segments, keywords/topics in Display), and the system can go beyond those signals to find additional people likely to convert.

Audience expansion is the lookalike-style option designed for reach and consideration in Video campaigns. When you enable it, Google Ads expands to people who are similar to the audience segments you selected—typically increasing reach while keeping the bidding model (like CPV/CPM) consistent with your setup.

Step 1: Build a strong “seed” audience (this determines lookalike quality)

Lookalikes are only as good as the source audience. In Google Ads, your best “seed” options usually come from two places: your website/app engagement (“your data” segments) and your customer list (Customer Match). I recommend you build both whenever possible, then test which one produces the best incremental new-customer performance.

Create a website visitor segment (“your data”)

This is the classic remarketing foundation. You’ll need the Google tag implemented so visitors can be added to your segments. Once your tagging is in place, you can create website-based segments in Audience Manager and define who qualifies using rules (for example: visited a product page, started checkout, or hit a specific URL pattern).

When creating the segment, pay special attention to membership duration. It should reflect your buying cycle. Shorter durations stay more “intent fresh,” while longer durations build volume (but may dilute intent).

Upload a Customer Match list (your highest-quality seed when it’s compliant and fresh)

Customer Match lets you target (and model from) people who have shared data with you, such as email addresses. From a “lookalike” standpoint, this is often your best seed because it represents real customers—not just anonymous site traffic.

Operationally, plan around a few realities: Customer Match lists can take up to 48 hours to process after upload, list memberships have a maximum 540-day membership duration, and you need ongoing refreshes to keep the list eligible and effective. If you set it and forget it, it will slowly decay.

Minimum sizes that matter (to avoid “no/low volume” headaches)

Audience eligibility and serving thresholds vary by network, and “estimated size” isn’t the same as “active users available to serve ads right now.” As a practical baseline, build seeds large enough that they can deliver consistently without constant learning resets.

• Website/app “your data” segments: aim to exceed the minimum active-user requirements for the networks you’ll advertise on (for example, Search and YouTube generally require more volume than Display).
• Customer Match: while lists can run with smaller minimums than typical remarketing on some surfaces, you’ll usually want well above 100 matched/active users for stable delivery—especially once you add any additional targeting restrictions.

Step 2: Turn that seed into “lookalike” reach (what to enable in campaigns)

Once you have a seed, your job is to choose the campaign type that matches your goal and then enable the correct expansion mechanism. This is where most advertisers get tripped up—because they try to force a “lookalike” concept into Search, when Google’s lookalike-style expansion is mainly built for Display and Video-style inventory.

Option A: Use Optimized Targeting (best for conversions on Display, Demand Gen, and eligible Video campaigns)

Optimized targeting is available for Display, Demand Gen, and Video action-style setups (goal-based video campaigns focused on sales/leads/traffic). It’s also effectively always part of Performance Max behavior, where audience signals guide automation but don’t hard-limit reach.

Conceptually, think of optimized targeting as: “Start with my seed… then find more people who behave like converters.” This is much closer to a conversion-optimized lookalike than a pure similarity model.

1. Go to the Ad group you want to use for prospecting.
2. Open Ad group settings and locate the Optimized targeting section.
3. Turn optimized targeting on (or confirm it’s on), then add your signals such as Customer Match and your website visitor segments. In Display, you can also use contextual signals like keywords/topics.
4. Keep exclusions clean: if you want truly new customers, exclude existing customer lists and/or key remarketing segments (and remember that exclusions can reduce volume and learning speed).

One important nuance: optimized targeting can serve beyond your selected signals. That’s the point. Your signals guide the model, but they aren’t a hard fence unless you’re using specific targeting modes that explicitly restrict delivery.

Option B: Use Audience Expansion (best for YouTube reach/awareness and consideration)

If your objective is Brand awareness and reach or Product and brand consideration, audience expansion is the cleanest “lookalike-style” switch for Video campaigns. You choose audience segments you like, and then enable audience expansion to reach more people who look similar to that selected audience.

1. Create a new Video campaign (or edit an existing one) built around awareness/reach or consideration.
2. In the People section (audience selection), pick the audience segments you want as your seed signals.
3. Check the Audience expansion box to enable expansion.
4. Monitor impact using the reporting row that combines expansion/optimized targeting activity, and evaluate it based on contribution to your overall campaign goal (reach, views, lift, assisted conversions, etc.).

If you’re running Video with Sales, Leads, or Website traffic goals, you’ll typically be steered toward optimized targeting instead of audience expansion—because the system is prioritizing conversion likelihood, not just similarity and reach.

What about Search campaigns—can Search do lookalikes?

Search doesn’t operate like social lookalikes. You can’t say, “Find people similar to my customers,” and have Search expand queries the way Meta expands users. Search is primarily intent-driven (keywords and query matching), but you can use your seed audiences in two smart ways.

First, use audiences in Observation mode so you don’t restrict reach, but you can see performance by audience and allow Smart Bidding to use those first-party audiences as signals. Second, if you want a strict “only show to this audience” approach (classic RLSA behavior), run a dedicated campaign/ad group using Targeting mode—just understand that this is not a lookalike; it’s a constraint.

Critical troubleshooting checklist (when your “lookalike” doesn’t spend or doesn’t work)

• List is too small or not “active” enough: delivery can fail even when the total list size looks fine, because only a portion of users are eligible/active at serve-time.
• Too many layers of restrictions: piling on tight geography, narrow demographics, aggressive exclusions, plus a small seed is the fastest way to “no/low volume.”
• Seed quality is mixed: if your seed includes low-intent visitors (like blog readers), your expansion will mirror that. Build intent-tiered segments (product viewers vs. cart starters vs. purchasers) and test them.
• Conversion signal is weak: optimized targeting learns from conversion behavior. If conversion tracking is incomplete or you’re optimizing to a low-quality conversion action, expansion will chase the wrong outcome.

The simple “expert rule” for better lookalike-style performance

If you want the closest thing to a high-performing lookalike in Google Ads, start with a clean customer list (or a high-intent website segment), use it as an audience signal, and let the system expand via optimized targeting (for performance) or audience expansion (for reach). Then measure success based on incrementality—especially if you’re excluding existing customers and trying to acquire new ones—rather than expecting a perfect 1:1 replacement for social lookalike mechanics.

‍

In Google Ads, there isn’t a classic “lookalike audience” button anymore; instead, the closest equivalent is building strong first-party “seed” audiences (like website/app data segments and Customer Match lists) and letting Google expand from them via optimized targeting (for performance-focused Display, Demand Gen, and conversion-oriented Video) or audience expansion (for reach-focused Video), while keeping an eye on list freshness, active user counts, and clean conversion signals so the automation can learn effectively. If you want a simpler way to keep this kind of setup healthy over time, Blobr connects to your Google Ads account and uses specialized AI agents to continuously analyze what’s working, flag audience and targeting issues that can limit scale, and turn best practices into clear, prioritized actions you can review and apply on your terms.