Lookalike Audience
Paid MediaAlso: LAL Audience · Similar Audience · Lookalike Segment
Quick definition
A lookalike audience is a targeting tool used in paid advertising that finds new people who share characteristics with an existing group you define. You supply a seed audience of real customers or high-value users, and the ad platform builds a larger audience of strangers who resemble them.
How it varies across Australia
Lookalike performance varies sharply by seed quality and market size. Australian advertisers face a structural disadvantage: the local addressable pool is smaller than US pools, which means tighter lookalikes become exhausted faster and broader lookalikes dilute quickly.
See acquisition performance patterns across Australian industries →The seed audience shapes everything
The source list you upload. High-value customers, recent converters, or specific event completions. Quality here determines everything downstream.
How broadly the platform casts. One percent is the tightest match. Ten percent is the widest. Smaller is more precise, not always more efficient.
The behavioural and demographic data the platform uses to match. Differs by platform. Meta uses on-platform and (historically) off-platform behaviour.
How recent the customers or events in your seed are. Older seeds reflect who your customers used to be, not who they are now.
What it actually means
Think of a lookalike audience as a recommendation from a very large, data-rich matchmaker. You hand the platform a list of your best customers. The platform looks at what they have in common across hundreds of behavioural and demographic signals, then finds other users on the platform who share those patterns. You never see the underlying signals. You just get an audience to target.
The seed audience is everything. If you feed the platform your full customer list, which includes your best and worst customers blended together, the signal is muddied. The platform has no way to know which behaviour to replicate. The most effective seeds are tight and intentional: your top-revenue cohort, repeat purchasers, customers who signed up in the last 90 days, or users who completed a high-value in-app action.
The size slider (on Meta it runs from 1 to 10 percent of the country) is a precision dial, not a volume dial. One percent means the platform finds the people who most closely resemble your seed. Ten percent means it casts wide enough to find anyone who vaguely fits. Smaller is more precise. Larger is more reach. They rarely perform the same.
Meta's version (Meta Lookalike Audiences) is the most widely used. Google's version (Similar Segments) was phased out for most campaign types in 2023. TikTok, LinkedIn and Pinterest all have equivalents. Each platform's model reflects its own data, which is why a Meta lookalike and a TikTok lookalike built from the same seed will target different people.
A lookalike audience is only as smart as the seed you give it. Feed it your average customers and you get more average customers.
How it shows up
Lookalike audiences show up in the audience targeting section of most major ad platforms. On Meta Ads Manager, you create them under Audiences by uploading a customer list, connecting a pixel event, or pointing to a page of people who interacted with your content. The platform then generates the lookalike and gives you an estimated reach number.
In campaign reporting, lookalike audiences appear as one audience segment alongside interest-based and retargeting audiences. The performance difference between a tight lookalike (small percent) and a broad lookalike (large percent) is one of the most telling signals about how well your seed audience reflects genuine customer quality.
The Australian context
Australia's population creates a ceiling on how large any lookalike can get before the quality drops. A one-percent lookalike of the Australian population is a fraction of the equivalent US pool, which means Australian advertisers reach exhaustion faster on small-percentage lookalikes and face tougher choices about whether to widen the percentage or shift to broader interest targeting.
Australian privacy obligations under the Privacy Act also affect how first-party customer data can be uploaded to ad platforms. Customer lists used as seeds for lookalikes should go through a consent and data-handling review before upload. Hashed email matching is the standard, but the underlying consent for use of that data in third-party advertising needs to be covered.
Where people get this wrong
Related terms
Common questions
What should I use as my seed audience?
Your best customers, not your full customer list. Filter for high-value purchasers, repeat buyers, or users who completed a meaningful action like a subscription renewal. The more specific and valuable the seed, the more useful the lookalike the platform builds.
Do lookalike audiences still work after iOS privacy changes?
They still run and still produce results, but the signal quality on Meta degraded after Apple's App Tracking Transparency (ATT) changes reduced the platform's visibility into post-click behaviour. Uploading strong first-party customer lists directly compensates for some of that signal loss.
How is a lookalike audience different from an interest-based audience?
Interest-based targeting uses declared or inferred interests like job title, hobby or content consumption. Lookalike targeting uses behavioural similarity to real customers you define. Lookalikes tend to perform better for conversion-focused campaigns when the seed is high quality.
How often should I refresh my lookalike seed?
Every four to six weeks for active campaigns. Customer quality and behaviour shifts over time. A seed built from last year's customer cohort may reflect a different type of buyer than you are acquiring now. Refreshing regularly keeps the lookalike anchored to your current best customers.
Keep exploring
About New Rebellion
New Rebellion is a marketing intelligence consultancy. We build tools, score Australian businesses on how their marketing actually performs, and publish Debrief every day. This dictionary is part of how we work in the open.
How we think →