Question 1

What is the best seed audience for Meta lookalike audiences?

Accepted Answer

The best seed audience for Meta lookalikes is your actual customer list - specifically verified purchasers from your CRM or email list. A customer list of 1,000 or more people gives Meta the clearest signal of who your buyers actually are. If you do not have a customer list, a pixel audience of 1,000 or more purchase events is nearly as strong. The key principle is that your seed should contain people who completed your desired action, not just visited your site. Value-based lookalikes, where you pass purchase value to Meta, are even more powerful when you have enough revenue data because they weight your best customers more heavily in the model.

Question 2

Should I use a 1% or 2% lookalike audience on Meta?

Accepted Answer

Start with a 1% lookalike audience when you are optimizing for conversions and efficiency. A 1% LAL contains the people most statistically similar to your seed - smaller but higher quality. Use 2-3% lookalikes when you need more scale and have a profitable 1% LAL already. Use 5-7% lookalikes for brand awareness campaigns or when your 1% audience is burning through too fast. The most common mistake is jumping straight to 5-10% LALs before finding a winning combination - this trades precision for scale before you know what works.

Question 3

How many people do I need in my seed audience for Meta lookalikes?

Accepted Answer

Meta requires a minimum of 100 people to create a lookalike audience, but the practical minimum for reliable results is 1,000 people. Below 1,000, Meta does not have enough data to build a dependable model and performance tends to be erratic. The ideal range is 1,000 to 50,000 people in your seed - large enough to give Meta signal but focused enough on your best customers. Above 50,000, you risk diluting the seed with lower-quality matches. For value-based lookalikes, you need at least 100 purchases with revenue values, but 1,000 is strongly recommended.

Question 4

How often should I refresh my Meta lookalike audiences?

Accepted Answer

Meta automatically refreshes pixel-based lookalike audiences every 3 to 7 days as new events come in. Customer list-based lookalikes are static and do not update automatically - you should re-upload your list and recreate list-based LALs every 30 to 90 days to incorporate new buyers. If your list is growing quickly (more than 20% per month), refresh monthly. The bigger threat to LAL freshness is pixel signal quality - iOS privacy changes reduce the number of events Meta can observe, which degrades purchase-based LALs over time. Check your pixel event match quality score in Meta Events Manager every month.

Question 5

Why are my Meta lookalike audiences not performing?

Accepted Answer

The most common reasons Meta lookalike audiences underperform are: seed audience is too small (under 1,000 people), seed quality is low (built from page views instead of purchases), too many LAL audiences running simultaneously and competing against each other, audiences not being properly excluded (retargeting overlapping with cold LAL campaigns), or the creative and offer are mismatched for a cold audience. A lookalike audience is only as good as its seed. If your best customers are not in the seed, Meta cannot build a model to find more of them.

Meta Lookalike Audience Planner

Related reading and tools