Lookalike Audiences

What This Page Answers

Lookalike audiences use a source audience to help platforms find people who resemble valuable existing users or customers. They are not magic prospecting audiences. They are only as good as the source audience, event quality, and business logic behind them. A lookalike built from bad leads will find more people like bad leads.

The Core Principle

Source quality beats source size until the source becomes too small to model. A large list of low-quality leads may create cheaper conversions, but it can teach the platform to find users who submit forms without buying. A smaller list of high-LTV customers may be strategically better, but if it is too small or stale, the platform may not have enough signal.

Strong Source Audiences

Good sources include:

Recent purchasers.
High-LTV customers.
Repeat buyers.
Qualified leads.
Closed-won customers.
Subscribers with retention.
High-value app users.
Customers from profitable product categories.

Weak sources include:

All site visitors.
Low-quality leads.
Giveaway participants.
Broad engagement audiences with no purchase signal.
Old customer lists with stale emails.
Mixed audiences where high-value and low-value users are not separated.

Lookalike Strategy Matrix

Source	Use Case	Risk
All purchasers	General acquisition	May ignore margin or LTV differences
High-value purchasers	Quality-focused acquisition	May be smaller and slower to learn
Qualified leads	B2B or lead gen prospecting	Requires clean CRM definitions
Repeat customers	Retention-like acquisition profile	May bias toward existing customer type only
Recent engagers	Early account exploration	Engagement may not predict purchase

How To Use Lookalikes

Use lookalikes to expand prospecting when:

First-party data is meaningful.
Broad targeting needs guidance.
You want to scale beyond exact retargeting.
You can compare against broad and interest audiences.
You can measure new customer quality.

Do not use lookalikes as a way to avoid creative strategy. The ad still needs to communicate who it is for and why it matters.

What To Measure

Compare lookalikes against broad targeting and interest hypotheses by:

CPA
CAC
New customer rate
Purchase value
LTV proxy
Lead quality
MER impact
Incrementality

If a lookalike has lower CPA but worse customer quality, it is not necessarily better.

Common Mistakes

Building lookalikes from all leads instead of qualified leads.
Using old customer lists without refreshing them.
Mixing geographies with different economics.
Comparing lookalikes to broad using different creative.
Treating lookalike performance as proof that the source audience is good.
Scaling lookalikes without checking new customer overlap.

Practical Rule

Lookalike quality follows source quality. Feed it the customers you want more of, not just the events you have the most of.