Paid Social
Lookalike Audience
A lookalike audience is a new group built by a platform to resemble an existing source audience, such as your customers or site visitors.
It helps find fresh prospects who share traits with people who already value your brand. For example, a lookalike of past purchasers targets users statistically similar to them. Source quality drives results, so a strong seed audience produces a stronger lookalike.
The seed decides everything: lookalikes model whoever you provide, so a seed of repeat purchasers produces buyer-like prospects, and a seed of everyone who ever clicked produces clicker-like ones. Value-based seeds go a step further, weighting by spend so the model chases your best customers, not your most numerous. Percentage is the reach dial: 1% is closest resemblance, 5 to 10% trades similarity for volume as budgets grow. Refresh seeds on a schedule, a lookalike of 2023's customers slowly stops describing the people 2026 should target.
Example
A cosmetics brand uploads its 2,000 repeat purchasers as a seed; the platform builds a 1% lookalike: 500,000 Thais who statistically resemble those buyers. Cost per first purchase: 240 baht against 610 for interest targeting. A second lookalike seeded from all site visitors, buyers and bouncers alike, performs no better than cold traffic: the seed's quality, not the mechanism, was the ingredient.
Frequently asked questions
- What seed audience makes the best lookalike?
- Your best customers, defined by value: repeat purchasers or a value-weighted list, at whatever size the platform needs to model from. Seed it with everyone who ever clicked and the model reproduces clicking, which is not the behaviour you wanted more of.
- What lookalike percentage should I use?
- Start at the tightest tier for the closest resemblance and highest expected conversion, then widen as budget and results allow. Broader tiers trade similarity for scale, so watch cost per result as you expand and stop where the economics stop.
Related terms
Working on Lookalike Audience? See how our Facebook Ads can help.
Explore Facebook Ads