Lookalike Audience: Meta vs Google Ads

Many digital marketers still talk about "lookalike audiences in Google Ads", even though that option no longer exists as such, since mid-2023. A lookalike audience is, on Meta Ads, a new audience generated automatically from a source audience (customers, website visitors, people who interacted with the Pixel), from which the platform identifies other users with similar traits and behaviors.

On Google Ads, the direct equivalent of this mechanism — called "similar audiences" — was fully removed in 2023, and its role was taken over by signal-based features such as optimized targeting and audience expansion. The two platforms now treat the concept of a "similar audience" in completely different ways, and the confusion between them is one of the most common misunderstandings in Google Ads and Meta Ads campaigns run by businesses in Romania.

This article explains the technical mechanism behind a lookalike audience on Meta, what changed in 2026 with Advantage+, what exactly happened to similar audiences in Google Ads, a numeric example with a RON budget, and the common mistakes that reduce the effectiveness of this audience type.

Expertise note: HappyWeb sets up lookalike audiences and Advantage+ campaigns in Meta Ads for businesses in Romania, integrated with each client's CPA and ROAS strategy, not as an option toggled on automatically, without a relevant data source.

How a lookalike audience works on Meta (Facebook and Instagram)

A lookalike audience is always built starting from a source audience — typically a Custom Audience made up of existing customers, website visitors (through the Meta Pixel), or people who interacted with the Page. Meta analyzes the shared characteristics of this source — interests, online behavior, demographics — and identifies, in the chosen country or region, other users with a similar profile.

The minimum source size accepted by Meta is 100 people, but a source that small gives the algorithm very few relevant signals. The practical recommendation, also used in accounts managed by HappyWeb, is a source of 1,000-5,000 people, built from an action with real value (a purchase, a qualified lead), not from a general audience without a clear shared behavior.

The similarity level: from 1% to 10%

When creating a lookalike audience, you choose a percentage between 1% and 10% of the target country's population. At 1%, the audience is closest to the source (high similarity, smaller reach); as the percentage grows toward 10%, the audience widens, but its similarity to the original source decreases. Meta automatically refreshes the audience's composition every 3-7 days, with no need to rebuild it manually.

This mechanism essentially automates an old principle from classic marketing: market segmentation, described by Philip Kotler in his work on marketing management as identifying groups of consumers with shared characteristics. A lookalike audience does exactly this through an algorithm, at scale, instead of manual market research — its role fits into the "Promotion" decision from E. Jerome McCarthy's 4P model (Product, Price, Place, Promotion, 1960): exactly who you direct your promotion budget toward.

From Lookalike Audience to Advantage+: what changed in 2026

In Advantage+ Shopping campaigns, Meta no longer treats a lookalike audience as a fixed target, but as a starting signal ("audience suggestion"): the system starts from the indicated source, but can expand delivery to other users if the algorithm estimates a higher conversion probability outside the initial segment. Some variants of this feature are labeled, in the interface, "Advantage+ lookalike" (formerly "lookalike expansion").

A manually built, fixed lookalike audience has not become useless — it remains relevant mainly at smaller budgets (under roughly $5,000/month) or when you have quality data from your own CRM that the Meta Pixel has never "seen". At larger budgets, with a steady volume of Pixel conversion events, Advantage+ tends to find relevant audiences automatically, without a manually built lookalike audience set up in advance.

Does a lookalike audience still exist in Google Ads? What happened to similar audiences

The short answer is no, not in the form it operated for years. Google officially announced, in November 2022, the gradual removal of "similar audiences" (also called "similar segments"). The process happened in two stages: starting May 1, 2023, Google Ads stopped generating new similar audiences for targeting and reporting (Search Ads 360 Help — Changes to audience targeting, checked 2026-06-25), and starting August 1, 2023, existing similar audiences were completely removed from all ad groups and active campaigns ( Google Ads Developer Blog — Announcing deprecation and sunset of Similar Audiences, checked 2026-06-25).

The official reason cited by Google was the shift toward advertising that relies less on third-party cookies, along with increasingly strict privacy requirements. In practice, any article or guide that still describes steps for "creating a similar audience in Google Ads" is describing a feature that no longer exists in the current interface.

What you use today in Google Ads instead of similar audiences

Google officially recommends two mechanisms, which work differently from the old lookalike model but pursue a similar goal — finding new, relevant users:

Optimized targeting — turned on automatically for most campaigns, it looks for relevant audience segments beyond the ones you selected manually, based on the conversion signals in your account (Google Ads Help — About optimized targeting, checked 2026-06-25).
Audience expansion — manually expands a chosen audience segment, looking for users with similar characteristics (Google Ads Help — Using audience expansion, checked 2026-06-25).
Customer Match — your own customer list (email, phone), used as a "signal" for Smart Bidding and optimized targeting, not as a lookalike audience in its own right (Google Ads Help — About Customer Match, checked 2026-06-25).

Google's official documentation points to optimized targeting as the closest substitute for the old similar audiences, with the recommendation to upload your own data segments (including Customer Match lists) as "hints" for the algorithm, instead of looking for a "lookalike" feature that, in Google Ads, no longer exists.

Meta vs Google Ads: where lookalike stands today, side by side

Criterion	Meta Ads	Google Ads
Dedicated "lookalike" feature	Available (Lookalike / Advantage+ lookalike)	Fully removed since August 2023
Minimum source size	100 people (1,000-5,000 recommended)	Not applicable (feature no longer exists)
Current expansion mechanism	Fixed lookalike or signal inside Advantage+	Optimized targeting + audience expansion
Role of first-party (CRM) lists	Direct source for the lookalike audience	Customer Match, used as a "hint" for the algorithm

Numeric example: how much a lookalike audience campaign costs, in RON

Illustrative example, to show the comparison logic: an online store in Romania runs, in parallel, a Meta Ads campaign on a 1% lookalike audience (source: customers who purchased in the last 6 months) and a campaign on a broad audience with no proprietary source, for one month, with equal budgets.

Audience type	Budget	Conversions	Resulting CPA
1% Lookalike (customer source)	3,000 RON	42	~71 RON
Broad audience, no source	3,000 RON	24	~125 RON

Illustrative figures, for demonstration purposes, showing the typical pattern observed in accounts managed by HappyWeb — a lower CPA on a lookalike audience when the source is high quality. They do not represent a real client's results and are not a universal benchmark; results vary by industry, source quality, and historical conversion volume.

Common mistakes with lookalike audiences and how to avoid them

Using a source that is too small or too general: a list of 100-200 people, without a clear shared action, gives the algorithm too few signals. Mitigation: build the source from an action with real value (a purchase, a qualified lead) and aim for 1,000+ people.
Not excluding the source audience from the lookalike campaign: you risk showing ads to people who are already customers, instead of reaching new users. Mitigation: explicitly exclude the source (and, ideally, recent converters) from the audience settings.
Automatically picking 10% "for more reach": a large percentage dilutes similarity to the source, especially if the source is already a niche one. Mitigation: test 1-3% first, especially for products/services with a clearly defined audience.
Looking for "similar audiences" in Google Ads, as in 2021-2022: the feature no longer exists since August 2023, and time spent searching for it is time wasted. Mitigation: set up optimized targeting and upload Customer Match lists as a signal, following Google's current documentation.
Never refreshing the lookalike source: an old source, based on customers from 1-2 years ago, no longer reflects the current profile of your best customers. Mitigation: review the source every 90-180 days, or whenever your offer changes significantly.

Frequently asked questions about lookalike audiences

What is a lookalike audience, in simple terms?

It is a new audience, automatically generated by Meta from a source audience (customers, website visitors), made up of other users with traits and behaviors similar to that source.

How large does the source need to be for a lookalike audience on Meta?

The minimum accepted by Meta is 100 people, but the practical recommendation is a source of 1,000-5,000 people, built from an action with real value, so the algorithm has enough relevant signals.

Do lookalike or similar audiences still exist in Google Ads?

No. Google fully removed "similar audiences" from Google Ads in two stages, in 2023: no new audiences starting May 1, and complete removal from campaigns and reporting starting August 1. The current substitutes are optimized targeting, audience expansion, and Customer Match.

What is the difference between a 1% and a 10% lookalike audience?

At 1%, the audience is closest to the source, with high similarity and smaller reach. As the percentage grows toward 10%, the audience widens, but its similarity to the original source decreases.

Conclusion: lookalike audience remains a Meta tool, not a universal one

A lookalike audience works well when the data source is high quality — a relatively small group with a clear shared action, not a large, general list. On Meta, the concept remains active and relevant, even though Advantage+ has increasingly turned it into a starting signal rather than a fixed audience. In Google Ads, the equivalent feature no longer exists since 2023, and confusing the two platforms is the most common strategy mistake tied to this term.

If you want lookalike audiences and Advantage+ campaigns built on a real data source, not assumptions, contact us for a free audit.

Want lookalike audiences and Advantage+ campaigns built right, from the first source?

HappyWeb sets up lookalike audiences, Advantage+, and Meta Ads (Facebook and Instagram) campaigns, as well as Google Ads campaigns with current targeting signals, not features that no longer exist. Contact us for a free audit.

Sources

Article last updated: 2026-06-25 · Recommended review: within 90-180 days, since Meta and Google frequently update their signal-based and automated targeting mechanisms.

Meta Business Help Center — "About Lookalike Audiences". Official documentation: facebook.com/business/help. Checked 2026-06-25.
Google Ads Developer Blog — "Announcing deprecation and sunset of Similar Audiences". Official documentation: ads-developers.googleblog.com. Checked 2026-06-25.
Search Ads 360 Help — "Changes to audience targeting: Google Ads will no longer support similar audiences". Official documentation: support.google.com/sa360. Checked 2026-06-25.
Google Ads Help — "About optimized targeting". Official documentation: support.google.com/google-ads. Checked 2026-06-25.
Google Ads Help — "Using audience expansion". Official documentation: support.google.com/google-ads. Checked 2026-06-25.
Google Ads Help — "About Customer Match". Official documentation: support.google.com/google-ads. Checked 2026-06-25.
E. Jerome McCarthy — Basic Marketing: A Global-Managerial Approach, for the 4P model (Product, Price, Place, Promotion, 1960) and placing targeting decisions within the "Promotion" area. Consulted from HappyWeb's internal library.
Philip Kotler — Marketing Management, for the market segmentation principle based on shared consumer characteristics. Consulted from HappyWeb's internal library.

If you have questions or want a tailored review of the audiences in your ad accounts, contact us.

Image generated with AI, used for illustrative purposes.

Lookalike Audience: How It Works in Meta and Google Ads