How to Create Facebook Lookalike Audiences That Find Your Best Customers

Lookalike Audiences are Facebook's most powerful targeting tool, and most advertisers either don't use them or use them wrong. When set up properly, a Lookalike Audience can outperform interest-based targeting by 200-400%, finding new customers who genuinely look like your best ones.

Here's how to create Lookalike Audiences that consistently outperform every other audience in your account.

What Lookalike Audiences Actually Do

A Lookalike Audience is exactly what it sounds like: Facebook finds people who share characteristics with a source audience you provide. If you upload a list of your best customers, Facebook will find more people who look like them — same demographics, interests, behaviours, and buying patterns.

The key word is "behaviours." Facebook's lookalike algorithm doesn't just match demographics. It models the actual behavioural patterns of your source audience and finds people with similar patterns.

This is why Lookalikes consistently outperform manual targeting. You'd never think to target "35-year-old women in Manchester who like yoga AND skincare AND Audi cars," but if that's the pattern in your customer base, Facebook will find more of them.

The Quality of Your Source Audience Determines Everything

This is the rule that almost every advertiser breaks: a Lookalike Audience is only as good as the source audience you give it.

If your source is bad, your Lookalike will be bad. If your source is small or unrepresentative, your Lookalike will be too. Spend time getting the source right and the Lookalike will work.

Source audience options, ranked from best to worst:

Highest-value customers (top 10-20% by lifetime value)
All paying customers
People who completed specific high-intent actions (purchased, booked, called)
Email subscribers who've engaged
Website visitors (broad version)
Page engagers (least specific)

Always start with #1 if you have the data. The difference between a Lookalike from your top customers vs all your customers can be 100-200% in performance.

Step 1: Create a Source Custom Audience

Go to Ads Manager → Audiences → Create Audience → Custom Audience.

Choose your data source:

For customer list (best option):

Click "Customer list"

Upload a CSV with email addresses (and phone numbers if available)

Facebook will hash and match these to user profiles

A list of 1,000+ matched customers makes a strong source

For website visitors:

Click "Website"

Choose "All website visitors" or specific pages (like the thank-you page)

Use the longest time window available (180 days)

For app activity, pixel events, video views, page engagement:

Each of these can also serve as a source

Best for businesses without an email list

Step 2: Build the Lookalike Audience

Once your source is created, go to Audiences → Create Audience → Lookalike Audience.

Settings:

Source: Select the Custom Audience you just created.

Location: Choose the country where you're running ads. You can do multiple countries if needed.

Audience size: This is where most people get it wrong. The slider goes from 1% to 10%.

1% = the closest match to your source. Highest quality, smallest audience.
3% = good balance for most businesses
5-10% = much broader, lower quality but more reach

Always start with 1%. It's tempting to go higher to get more reach, but quality beats quantity. A 1% Lookalike of 250,000 people is plenty for most campaigns.

Step 3: Test Multiple Lookalike Tiers

Don't put all your eggs in one Lookalike. Create several at different sizes and test them:

1% Lookalike of top customers
2% Lookalike of top customers
3% Lookalike of all customers
1% Lookalike of website conversion events

Run each in a separate Ad Set with the same budget and creative. After a week, you'll see which one performs best for your business.

Step 4: Don't Stack Interests on Top of Lookalikes

This is the single most common mistake. People create a Lookalike Audience and then layer in interests on top to "refine" it. Don't do this.

Lookalikes already incorporate the demographics, interests, and behaviours of your source audience. Adding more layers tells Facebook you're second-guessing the algorithm and shrinks the audience unnecessarily.

Run your Lookalike clean. Let Facebook's machine learning do its job.

The one exception: location targeting. Adding a geographic restriction is fine because it's about reachability, not behaviour.

Step 5: Use Lookalikes at the Right Stage

Lookalikes are cold audience tools. They're for prospecting — finding new people who don't know your brand yet. Use them at the top of your funnel.

For your funnel structure:

Top of funnel (cold): Lookalike Audiences

Middle of funnel (warm): Custom Audiences (page engagers, video viewers)

Bottom of funnel (hot): Website visitors and cart abandoners

Don't waste a good Lookalike on retargeting audiences. Use it where it has the biggest impact: finding new customers.

Step 6: Refresh Your Source Audiences

Lookalike Audiences automatically refresh every 3-7 days based on the source. But the source data needs to stay current.

Customer list: Re-upload monthly with new customers added
Website pixels: Update automatically
Engagement audiences: Update automatically

An outdated source produces a stale Lookalike. Stay on top of this.

Step 7: Combine Sources for Better Quality

Facebook lets you create a Lookalike from up to 10 source audiences combined. This is genuinely useful.

For example, you could create a Lookalike from:

Customer purchase list

People who watched 75% of your video ads

Engaged Instagram followers

This combined source gives Facebook a much richer behavioural signal than any single source alone.

Real-World Lookalike Strategies

E-commerce (with strong customer data):

1% Lookalike of customers who've purchased twice or more

Run as your primary cold audience

Expect 2-3x better ROAS than interest-based targeting

Local service business (gym, salon, restaurant):

Customer list + page engagers

1-2% Lookalike with location restriction (10-mile radius)

Often outperforms broad interest targeting by 50-100%

Coach/consultant (small customer list):

1% Lookalike of email subscribers who've opened multiple emails

1% Lookalike of people who watched a webinar

Layer with broad interest categories for scale

B2B service:

1% Lookalike of customers from a specific industry

2% Lookalike of demo bookers

Critical: keep source quality high

Common Lookalike Mistakes

Source audience too small. Facebook needs at least 100 people for a Lookalike, but performance dramatically improves with 1,000+. Below 500, results are unpredictable.

Source audience too broad. A Lookalike of "all website visitors in the last 6 months" includes people who bounced after 2 seconds. Filter to higher-intent visitors.

Going too big too fast. Always start with 1%. You can scale to larger sizes once you've validated that your source produces good Lookalikes.

Stacking interests on top. Defeats the purpose. Run Lookalikes clean.

Not testing multiple sources. Different source audiences produce different Lookalikes. Test them.

Forgetting to refresh. Stale sources produce stale Lookalikes.

Performance Benchmarks

Good Lookalike Audiences typically achieve:

CTR: 1-2.5% (similar to good interest targeting)
CPC: Often 20-30% lower than interest targeting
ROAS: 2-4x higher than interest targeting for the same budget
Conversion rate: Often 50-100% higher than interest audiences

If your Lookalike isn't outperforming your interest-based ads, your source audience needs work.

Scaling Lookalike Audiences

Once you have a winning Lookalike:

Scale budget gradually (20% increases every 2-3 days)
Test additional Lookalike sizes (move from 1% to 2% to 3%)
Test new source audiences for the same product/offer
Combine top-performing sources into multi-source Lookalikes

The Honest Limitation

Lookalike Audiences only work if you have meaningful data to start with. If you're a brand new business with no customers, no email list, and no website traffic, you can't use Lookalikes effectively. Start with broad interest-based targeting until you've generated enough source data (at least 100 customers or 500 website conversion events).

After that, switch to Lookalikes as your primary cold audience strategy. The performance gap is real and consistent.

Managing multiple Lookalike tests, refreshing source data, and continually optimising audience performance is genuinely complex. Pix-Vu handles Lookalike Audience strategy and optimisation as part of its £99/month Facebook ad management — letting you benefit from the algorithm without becoming an audience expert yourself.

Lookalikes are the closest thing Facebook ads has to a cheat code. If you have the source data, use them.