Lookalike Audiences in 2028: Do They Still Work?

Four years after iOS 14 reshaped digital advertising, the question of whether lookalike audiences still work keeps coming up. The short answer: they work differently than they used to, and most brands are using them wrong.

The long answer requires understanding what changed, what Meta's algorithm actually does now, and how to build audience strategies that work with the current system instead of fighting it.

What Changed (And Why Old Playbooks Fail)

Pre-iOS 14, Meta had near-perfect data on user behavior across the internet. The Facebook pixel tracked everything — every page view, every add-to-cart, every purchase, even across other websites. Lookalike audiences built from this data were incredibly accurate because Meta could match your buyers against detailed behavior profiles of billions of users.

Post-iOS 14, roughly 70-80% of iPhone users opted out of cross-app tracking. Meta lost visibility into what those users do outside of Facebook and Instagram. The pixel still works for users who opted in, but the dataset it builds from is significantly smaller and less reliable.

The result:

• Lookalike audiences based on pixel events (purchases, add-to-carts) are less accurate than they were
• Attribution is muddier — you can't always tell which ad drove the purchase
• Performance has become more volatile and harder to predict

But Meta didn't just accept this decline. They rebuilt their ad system around it. And that's where the opportunity is.

How Meta's System Works Now

Meta responded to the data loss by leaning hard into two things: machine learning optimization and on-platform data.

Advantage+ campaigns essentially let Meta's algorithm decide who sees your ads, how much to bid, and which creative to show. Instead of you defining the audience, you give Meta a broad targeting directive and strong creative, and the algorithm figures out the rest.

Conversions API (CAPI) is Meta's server-side tracking solution. It sends conversion data directly from your server to Meta, bypassing browser-based tracking limitations. This gives Meta more accurate conversion data to optimize against, partially offsetting the iOS-related data loss.

On-platform signals — likes, comments, saves, video views, profile visits, and messaging — are unaffected by iOS changes. Meta now weights these signals more heavily in audience building and ad optimization.

The brands that are winning in 2028 understand this shift and have adapted their audience strategy accordingly.

Lookalike Audiences: What Still Works

1. Lookalikes From Customer Lists (Best Performance)

The strongest lookalike audiences are now built from first-party data — specifically, your customer email lists synced to Meta via Klaviyo or uploaded directly.

Why they work: This data isn't affected by iOS tracking. You're giving Meta a list of actual buyers, and Meta matches them against its own dataset of 3+ billion users. The matching happens on Meta's side using deterministic data (email, phone number), which is more reliable than probabilistic pixel matching.

Best source lists for lookalikes:

• All purchasers (last 180 days): Your broadest and most reliable seed audience
• Repeat purchasers: Customers who bought 2+ times. These are your best customers — a lookalike of repeat purchasers finds more people like your best people.
• High-LTV customers: Segment your top 20% by total spend. A lookalike of these customers is optimized for value, not just volume.
• Recent purchasers (last 30 days): Fresher data = more relevant signal for Meta's algorithm.

How to set them up:

1. Sync your Klaviyo segments to Meta (Klaviyo has a built-in integration for this)
2. In Meta Ads Manager, create a Custom Audience from the synced list
3. Create a Lookalike Audience from that Custom Audience
4. Test 1%, 3%, and 5% lookalike sizes

Our recommendation: Start with 1% lookalike of all purchasers (last 180 days) as your core prospecting audience. Layer 3% and 5% as you scale.

2. Lookalikes From Engagement Audiences

On-platform engagement isn't affected by iOS changes. Meta knows exactly who interacted with your content on Facebook and Instagram.

Engagement audiences that work for lookalikes:

• Video viewers (watched 50% or more of a video ad): These people showed genuine interest
• Instagram profile visitors (last 90 days): Active brand interest
• People who engaged with ads (liked, commented, saved, shared)
• People who messaged your page

Why these work: Someone who watched 75% of your product demo video is demonstrating interest that Meta can reliably track. A lookalike of these high-engagement users finds similar people who are likely to engage similarly.

3. Lookalikes From Conversions API Data

If you have Conversions API set up (and you should), Meta receives server-side purchase data that's more complete than pixel-only data.

Build lookalikes from CAPI-reported events:

• Purchase events (most reliable)
• Add-to-cart events
• Initiate checkout events

CAPI data fills in the gaps that pixel data misses, giving Meta a more complete picture of who your buyers are.

What Doesn't Work Anymore

Tiny Niche Lookalikes

Pre-iOS 14, you could build a lookalike from a list of 100 customers who bought a specific product and get solid results. Now, small seed audiences produce unreliable lookalikes because Meta doesn't have enough data points to find meaningful patterns.

Minimum seed audience size: 1,000 for decent results. 5,000+ for strong performance. Below 1,000, your lookalike will be barely distinguishable from broad targeting.

Pixel-Only Event Lookalikes (Without CAPI)

If you're still relying solely on the Meta pixel for event tracking, your conversion data is incomplete. The pixel misses a significant percentage of conversions on iOS devices. Your "purchase" lookalike audience is built from an incomplete and biased sample.

Fix: Install Conversions API. Shopify has a native integration. It takes 15 minutes and immediately improves both your lookalike quality and your campaign optimization.

Stacking Multiple Narrowing Interests on Top of Lookalikes

The old trick of taking a 1% lookalike and then narrowing it by "interested in organic skincare AND yoga AND Whole Foods" worked when Meta had deep behavioral data. Now, narrowing reduces your audience size without meaningfully improving quality, because Meta's interest targeting is less accurate than it used to be.

Better approach: Use broad lookalikes (1-3%) and let Meta's delivery optimization do the narrowing for you. Meta's algorithm is better at finding your buyer within a 1% lookalike than you are at defining them through interest filters.

The Advantage+ Question

Meta wants you to use Advantage+ Shopping campaigns instead of traditional campaign structures. In Advantage+ campaigns, you don't set audience targeting at all — Meta finds the audience for you.

Should you abandon lookalikes entirely for Advantage+?

Based on what we see across the brands we manage: no. But you should test both.

When Advantage+ outperforms lookalikes:

• Brands spending $10K+/month on Meta (enough data for Meta to optimize effectively)
• Brands with broad appeal products (not hyper-niche)
• Brands with strong, diverse creative (Advantage+ needs creative variety to test)
• Brands with CAPI installed (better data = better algorithmic optimization)

When lookalikes outperform Advantage+:

• Brands spending less than $5K/month (not enough data for Advantage+ to learn)
• Niche products with specific buyer profiles
• New brands with limited creative assets
• Brands without CAPI installed

Our approach: Run both. Allocate 50-60% of budget to Advantage+ and 40-50% to traditional campaigns with lookalike audiences. Compare performance over 30-day windows. Shift budget toward the winner, but don't kill the loser entirely — what works today might not work next month.

The Creative-Is-Targeting Reality

Here's the uncomfortable truth about 2028 Meta advertising: your creative is your targeting now.

When audience targeting was precise, mediocre creative could perform because it was shown to exactly the right people. Now that targeting is broader and more algorithmic, the creative itself determines who engages.

A UGC video of a 28-year-old woman talking about your skincare product will naturally attract 25-35 year old women interested in skincare — not because you targeted that demographic, but because Meta shows the ad to people most likely to engage with it, and that content resonates with that audience.

Practical implication: Instead of building 10 audience segments with specific targeting, build 10 creative variations that speak to different customer personas. Launch them all in a broad or Advantage+ campaign and let Meta's algorithm match each creative to the right audience.

This sounds counterintuitive if you're used to the old playbook. But the data consistently supports it.

Building Your 2028 Audience Strategy

Here's the complete framework:

Layer 1: Advantage+ Shopping Campaign (50-60% of budget)

• Broad targeting, no exclusions
• 5-10 creative variations
• Let Meta optimize fully

Layer 2: Lookalike Campaigns (25-35% of budget)

• 1% lookalike of all purchasers (last 180 days) from customer list sync
• 1% lookalike of repeat purchasers
• 1% lookalike of high-engagement video viewers
• Test 3% and 5% as you scale

Layer 3: Retargeting (15-20% of budget)

• Website visitors (last 30 days) who didn't purchase
• Email subscribers who haven't purchased
• Cart abandoners (last 14 days)
• Video viewers (75%+) who haven't visited the site

Data infrastructure requirements:

• Conversions API installed and sending purchase events
• Klaviyo audience sync active (updating daily)
• UTM parameters on all ad links
• Minimum 1,000 customers in your seed list for lookalikes

The Bottom Line

Lookalike audiences aren't dead. They're different. The brands that adapted to the new reality — first-party data, Conversions API, broad plus algorithmic optimization, creative as targeting — are outperforming brands that are still trying to run the 2020 playbook.

The biggest mistake is doing nothing. If your Meta performance has declined and you haven't updated your audience strategy in the last year, you're leaving money on the table.

If you want us to audit your Meta ad account, rebuild your audience strategy, and set up the data infrastructure that makes it all work, book a call. We'll show you exactly where the gaps are and what to fix first.