Lookalike Audiences

Lookalike Audiences are modeled prospecting audiences that help ad platforms find new people similar to a source audience (e.g., your purchasers, high-LTV customers, or site engagers). On Meta, you choose a percentage of a country’s population (smaller % = more similar; larger % = broader reach). Meta can also expand beyond your selected size with Advantage+ lookalike to improve performance. On TikTok, you pick Narrow / Balanced / Broad lookalike sizes. 

You can seed lookalikes from pixel/app events, CRM lists, or engagement audiences; Meta also supports value-based lookalikes when your seed includes customer value (e.g., CLV or purchase value). Customer lists are hashed before upload per platform policy. 

Why It Matters

  • Scale what already works: Reach more people who resemble your best customers without guessing interests. 

  • Quality over volume: Starting with a small, high-quality seed (e.g., recent purchasers or high CLV) typically yields higher intent. 

  • Platform shift awareness: Google Ads removed Similar Audiences (its lookalike) and now uses optimized targeting/audience expansion; LinkedIn discontinued lookalikes in Feb 2024 (moving toward Predictive Audiences). Knowing where lookalikes still exist matters for planning. 

Examples

  • Meta: 1% Purchasers Lookalike: Seed = last 180-day purchasers; build 1% LAL for your target country; layer Advantage+ lookalike to allow expansion if it improves results. 

  • Meta: Value-Based Lookalike: Upload a customer list with value column (e.g., CLV or revenue). Create a value-based LAL to find more high-value prospects. 

  • TikTok: Narrow vs Broad: Use a Narrow lookalike (highest similarity) for efficiency tests and a Broad lookalike for scale; TikTok requires ≥1,000 matched users in the source Custom Audience.

Best Practices

  1. Use the right seed. Prioritize purchase, high-LTV, or qualified lead lists over generic site traffic. (That’s also how you unlock value-based LALs.) 

  2. Mind minimums & recency. For Meta, a practical rule is ≥100 people from one country (Meta commonly recommends 1k–5k high-quality seed). Keep seeds recent. 

  3. Test sizes. Start at 1% then test wider bands (e.g., 2–3%, 5–10%) or allow Advantage+ lookalike to expand. 

  4. Exclude customers in prospecting. Use exclusion targeting to avoid wasting budget on current customers when your goal is acquisition. 

  5. Compare against “broad/AI” options. Benchmark LALs vs Meta Advantage+ audience (broad with signals) and platform expansion/optimization features. 

  6. Stay compliant. Ensure proper consent and follow platform rules; customer data is hashed before upload.

Related Terms

  • Custom Audience / Customer List 

  • Value-Based Lookalike 

  • Advantage+ Lookalike 

  • Optimized Targeting / Audience Expansion (Google Ads) 

  • Predictive Audiences 

FAQs

Q1. How big should my source audience be?
For Meta, at least 100 people from a single country; many guides recommend 1k–5k for stability (quality matters more than size). TikTok requires ≥1,000 matched users in the source Custom Audience. 

Q2. What percent should I choose on Meta?
Start with 1% (most similar) for efficiency, then test 2–3% and 5–10% for scale. You can also enable Advantage+ lookalike to let the system expand when it predicts better results. 

Q3. Do lookalikes still exist on every platform?
No. Google Ads removed Similar Audiences (use optimized targeting/audience expansion instead). LinkedIn discontinued lookalikes on Feb 29, 2024 (use Predictive Audiences / Audience Expansion). Meta and TikTok still support lookalikes. 

Q4. Should I exclude existing customers?
Yes, when prospecting. Add exclusions (e.g., converters or customer lists) so budget focuses on net-new prospects. 

Q5. Do Meta lookalikes update over time?
Meta lookalikes typically refresh periodically while they’re in active use in campaigns. (Practitioner docs note an ~3–7 day refresh cadence.) 



Lead routing ›