Key Takeaways:
- Signal Loss Made Seeds Smaller And Noisier: Most iPhone users decline app tracking, so pixel-based seeds now miss many of the buyers they are meant to represent.
- Your Best Customers Make The Best Seed: Value-ranked customer lists and server-confirmed purchases point Meta toward profitable buyers instead of average ones.
- Lookalikes Now Guide Delivery Rather Than Limit It: Inside Advantage+, a lookalike works as a suggestion, so seed quality shapes results more than the percentage you choose.
Think of a lookalike audience as a photocopier. Meta reads the original you give it and prints as many copies as your budget allows, but every copy inherits the flaws of the source. After Apple's iOS privacy changes, the original brands rely on got thinner and less accurate, and their prospecting results have been printing faint ever since.
At Nord Media, we run Meta and TikTok campaigns as one part of a growth plan that also covers search, creative strategy, and the financial targets behind every channel. Audience work is where those pieces meet, because the customers a seed points Meta toward deciding how efficiently creative performs and how profitable each new customer turns out to be.
In this guide, we explain how iOS privacy changes weakened lookalike seeds, which data sources produce stronger ones, how to rebuild a richer seed with first-party data, and where lookalikes fit inside Meta's automated delivery today.
How iOS Privacy Changes Weakened Lookalike Seeds
The problem started with what Meta could no longer see.
Apple launched App Tracking Transparency with iOS 14.5 in April 2021, and global opt-in rates settled near 25%. Meta told investors in February 2022 that the change would cost roughly $10 billion in revenue that year. For advertisers, the practical effect was fewer matched events, which meant seed audiences built from pixel purchases or site visits shrank and lost many of the customers they were meant to describe.
A smaller seed is only part of the issue. When some buyers are missing and others are overrepresented because of how they browse, the seed stops reflecting your real customer base. Meta then searches for people who resemble a skewed sample, and prospecting costs climb while new customer quality slips.
The Seed Sources That Produce Stronger Lookalikes
Seed sources vary widely in how well they describe your ideal buyer, and the list below ranks them from most to least reliable.
- High-LTV Customer Lists: Customers ranked by lifetime spend give Meta the clearest view of who is worth acquiring, which makes this the strongest lookalike seed audience for most ecommerce brands.
- Repeat Purchasers: Buyers who return at least once show that the product fits their needs, which filters out one-time discount shoppers.
- Purchasers Of High-Margin Products: A seed built from customers of your most profitable lines steers prospecting toward the revenue that funds growth.
- Subscribers Who Converted: Email or SMS subscribers who later bought combine first-party contact data with proven intent.
- Engaged Viewers And Visitors: Video viewers and site visitors are useful for newer brands, though they reflect interest rather than buying behavior.
Moving up this list as your customer data grows gives Meta a sharper original to copy.
Building A Richer Seed Without Third-Party Tracking
First-party data now does the work that cross-app tracking used to do.
Send Purchase Data From Your Server
Browser-based tracking misses purchases when cookies are blocked or users decline tracking. Pairing the pixel with Meta CAPI sends confirmed purchases from your server along with hashed customer details, so more real buyers are matched and included in your seeds.
Add Value So Meta Learns Who Matters
A value-based lookalike uses a customer list or events that carry a value for each person, such as total spend or lifetime value. Meta then looks for people who resemble your most valuable customers rather than treating a one-time buyer and a loyal customer as equal.
Refresh And Clean The List On A Schedule
Customer lists age quickly as people change emails and buying habits. Updating seeds monthly, removing refunded or fraudulent orders, and including phone numbers and names alongside emails all help Meta match more of your list.
Putting Lookalikes To Work Inside Advantage+
Lookalikes still have a role in Meta's automated campaigns, but that role has changed.
Treat Lookalikes As Signals Instead Of Boundaries
In current campaign setups, Meta treats custom and lookalike audiences as audience suggestions, showing ads beyond them when it predicts better performance. Advantage+ lookalike, previously called lookalike expansion, works the same way. A strong seed still matters because it gives the system a better starting point.
Keep Prospecting And Retargeting Separate
A sound meta custom audience strategy gives each audience a clear job. Lookalikes are built to find new customers, while site visitors, cart abandoners, and past buyers need different messages, such as reminders, offers, or cross-sells, which is the role of Facebook retargeting. Keeping those jobs apart also keeps prospecting results honest: in manual campaigns, exclude recent purchasers from lookalike ad sets, and in Advantage+ sales campaigns, identify existing customers and cap the share of budget spent on them.
Test Lookalikes Against Broad Before You Scale
Much of the Facebook lookalike audience 2026 debate centers on whether broad Advantage+ delivery has made lookalikes unnecessary, and your own account is the only reliable place to settle it. Test a value-based lookalike against broad delivery with identical creative and budget, scale whichever brings in profitable new customers, and follow the same discipline that applies to scaling Facebook ads generally, where creative volume usually matters as much as the audience.
Final Thoughts
A meta lookalike audience is only as good as the seed behind it, and iOS privacy changes made that seed harder to build. Brands that send server-side purchase data, rank customers by value, and keep their lists clean give Meta a clearer original to copy, even as delivery becomes more automated.
The practical next step is an audit of your current seeds: where the data comes from, when it was last refreshed, and whether it reflects the customers you actually want more of. A lookalike that finds cheap customers who never return is a cost rather than a growth lever, so judge every seed by the value of the buyers it brings in.
Frequently Asked Questions About Meta Lookalike Audiences
What percentage should a lookalike audience use?
A 1% lookalike is the closest match to your seed, while larger percentages up to 10% trade similarity for reach.
How many people does a seed audience need?
Meta requires at least 100 people from one country, though larger and cleaner seeds give the model more to learn from.
Should existing customers be excluded from lookalike campaigns?
Yes, excluding current customers keeps prospecting budget focused on people who have not bought from you yet.
Are there limits on which custom audiences can be used?
Yes, since September 2025 Meta has more actively restricted custom audiences that could suggest sensitive information about people.
How large should a Meta prospecting audience be?
Meta recommends a size between two and ten million people for most audiences, since smaller audiences tend to cost more to reach.
How long does a new lookalike ad set take to stabilize?
Meta says ad sets typically exit the learning phase after about 50 optimization events within seven days.
.png){kind=tracking}
