A lookalike audience is a targeting tool offered by paid social platforms - primarily Meta and TikTok - that uses machine learning to find new users who share behavioural and demographic characteristics with a defined source audience. You provide the platform with a seed list (your best customers, your email subscribers, your highest-LTV purchasers), and the algorithm identifies patterns in that group's platform behaviour to find millions of similar users who have never interacted with your brand but statistically resemble people who have converted.
Lookalike audiences are one of the primary prospecting tools for Shopify brands running paid social. They sit between broad interest targeting (low precision, high reach) and retargeting (high precision, limited scale) - offering audience expansion beyond your existing customers while maintaining some relevance signal. For most DTC brands, a well-built lookalike of their top customers outperforms interest-based targeting in cost per acquisition.
The quality of a lookalike audience is determined entirely by the quality of the seed list. The algorithm can only find people similar to who you give it, so the seed should represent your best customers - not all customers. Common high-performing seed lists include:
High-LTV customer list: customers who have purchased 2+ times, or your top 20% by spend. This produces a lookalike biased toward repeat buyers rather than one-time purchasers.
Recent purchasers: customers acquired in the past 30-90 days. Fresher data reflects current purchase patterns more accurately than a years-old list.
Email subscribers with high engagement: a Klaviyo segment of subscribers who consistently open and click is a strong signal of brand affinity even before a purchase occurs.
Meta recommends a seed list of 1,000-50,000 people for optimal lookalike quality. Below 1,000 the signal is too thin; above 50,000 the precision dilutes. For Shopify brands with smaller customer bases, combining customer lists with high-engagement website visitors (via the Meta Pixel) can supplement the seed size.
Meta allows you to create lookalikes at 1% through 10% of a country's population. A 1% lookalike is the smallest audience but most closely resembles your seed - highest precision, highest conversion rates, highest CPMs. A 10% lookalike is much broader, with lower similarity to your seed but far greater reach and typically lower CPMs. Most Shopify brands test 1-3% lookalikes first, then expand to 3-5% or 5-10% as budgets scale and 1% audiences saturate.
Apple's ATT changes reduced the signal available from pixel-based events, making pixel-built lookalike audiences smaller and less precise. The most resilient lookalike seeds post-iOS 14 are first-party data uploads (customer email lists matched to Meta accounts) rather than pixel event audiences, since they are not affected by cookie-based tracking limitations. Brands that have built large, high-quality email lists through Klaviyo have a structural advantage in building effective lookalike audiences as third-party tracking continues to erode.
.png)
.png)
