Facebook Lookalike Audiences: Do They Still Work in 2026?

Yes — but not the way they did in 2018.

Pre-iOS 14, a 1% lookalike of your purchasers was the cheat code. Almost always outperformed broad targeting. Today the gap has narrowed and in many accounts, broad targeting beats lookalikes.

But lookalikes still have their place. Here's how to use them in 2026.

TL;DR

Use lookalikes when: source audience has at least 1,000 high-quality events.

Best source: 90-day Purchasers, 180-day high-value Purchasers, 30-day video viewers (75%+).

Best size: 1% for cold prospecting, 2–5% for scale, 6–10% for very large budgets.

Don't use lookalikes for retargeting (use custom audiences instead).

Test lookalikes vs broad in ABO. The winner gets your spend.

How lookalikes work

Meta takes your source audience (say, your 5,000 most recent purchasers) and finds 2 million Filipinos who share characteristics with them. That's your "1% lookalike."

In 2018, this was magic. In 2026, Meta's broad-audience algorithm has gotten so good that it often finds the same 2M people without your help.

But there are still situations where lookalikes outperform broad:

New ad accounts with limited Pixel data.

Niche product categories.

Brand-specific buyer profiles that don't fit broad demographics.

Source audiences (most important decision)

The lookalike is only as good as its seed.

Tier 1 sources (best for prospecting)

90-day Purchasers: most recent paying customers.

180-day Purchasers (especially if you're seasonal).

High-value Purchasers (top 25% AOV in last 180 days).

These are real buyers who paid you money. The strongest signal you can give Meta.

Tier 2 sources (good for warming up)

Subscribers/email list (uploaded as customer file).

Klaviyo "Engaged 90 days" (uploaded).

Loyalty program members.

Tier 3 sources (use cautiously)

30-day video viewers (75%+) — useful when Pixel data is limited.

Page engagers (lower-quality signal).

Add-to-cart but not purchase (medium signal).

Avoid

Bounces, brief PageView only audiences.

Audiences under 100 people (too small, low quality).

Anything you wouldn't model your perfect customer on.

Lookalike sizes

Meta lets you create 1%, 2%, 3%, ..., up to 10%. Smaller = more similar.

1% lookalike: most similar to source. Smallest audience (~2M in PH). Best CPA usually.

2–5%: still similar but bigger. Useful when scaling beyond 1%.

6–10%: looser. Closer to broad targeting. Use for big budgets.

For most accounts: start with 1%, expand to 2–3% as you scale, jump to 5–10% only at high spend.

Stacked lookalikes

You can combine multiple lookalikes in one audience:

1% Purchasers (90 days)

2% Purchasers (180 days)

1% AddToCart (90 days)

This creates a single larger audience that's still relatively similar to your buyer profile.

When to use: when 1% alone is too small for your budget but 5% is too loose.

Geographic considerations (PH-specific)

Lookalikes default to a single country. For PH stores:

Country: Philippines (default).

Language: usually doesn't matter for PH.

Region: optional, narrows audience.

If you also sell to Singapore or Malaysia, create separate per-country lookalikes.

How to test lookalikes vs broad

Run ABO with 4 ad sets at equal budget:

Ad Set A: 1% LAL Purchasers

Ad Set B: 2% LAL Purchasers

Ad Set C: 5% LAL Purchasers

Ad Set D: Broad (no audience targeting)

Each at ₱500/day for 7–14 days. Same creative across all.

Measure: CPA, ROAS, frequency. Pick the winner. Move winner to a CBO/ASC campaign.

Common results in 2026

Broad wins: 60% of accounts I test.

1% LAL wins: 20%.

2–5% LAL wins: 10%.

6–10% LAL wins: 10%.

But the gap is small. If broad wins by 5%, but lookalike provides faster scaling stability, lookalike may still be worth using.

Lookalike fatigue

Lookalikes shrink and stale over time:

Source audience refreshes too slowly → lookalike repeats.

Same lookalike running for 6+ months → frequency creeps.

Audience overlap with prior campaigns.

Refresh: rebuild your top lookalikes every 90 days from updated source audiences.

Custom audiences for retargeting (not lookalikes)

Don't use lookalikes for retargeting. Use custom audiences:

30-day website visitors.

30-day product viewers.

14-day add-to-cart, no purchase.

14-day initiated checkout, no purchase.

Lookalikes are for prospecting, not retargeting.

Lookalike combinations that work

For mid-funnel prospecting

1% LAL Purchasers + EXCLUDE recent purchasers and add-to-cart.

For scale

2% LAL Purchasers (180-day) + 1% LAL AddToCart (90-day).

For cold prospecting on a new pixel

1% LAL of your email list (uploaded customer file).

1% LAL of 75% video viewers from organic content.

Audience overlap

Run Meta's Audience Overlap tool monthly:

1% Purchasers vs 2% Purchasers will overlap heavily — don't run both.

1% Purchasers vs 1% AddToCart will overlap less.

Lookalike vs broad will have unpredictable overlap.

If overlap exceeds 30%, you're competing against yourself. Consolidate.

Privacy and tracking changes

Each year, lookalikes lose a bit of fidelity:

iOS 14.5 reduced signal.

iOS 17 added more privacy.

Cookie deprecation continues.

CAPI (Conversion API) is essential. Without server-side events, your source audiences shrink and lookalikes degrade faster.

When to skip lookalikes entirely

If you're running ASC, lookalikes don't apply — ASC handles audience automatically.

If your account is brand new (under 30 purchases), lookalikes don't have enough seed data. Run broad until you have a decent purchase history.

If your target ROAS is comfortable on broad, don't add complexity. Lookalikes are a tool, not a requirement.

Decision framework

1. Are you running ASC? → Skip lookalikes.

2. Brand new account, <30 purchases? → Skip lookalikes; run broad.

3. Established account, want to test? → Run ABO comparing 1% LAL vs broad.

4. Scaling beyond what broad alone can deliver? → Add 2–5% LAL as a layer.

5. Lookalikes underperforming broad? → Trust broad, scale broad.

Want help building audiences?

Audience strategy is one of the trickiest parts of media buying. My Facebook Ads Specialist service includes audience builds and testing. Or learn the system in the Facebook Ads Course Philippines.