Custom Audiences and Lookalikes From Scratch: The Complete Guide

Custom Audiences and Lookalikes are the two audience tools that make Meta advertising fundamentally different from cold, interest-based targeting, and most accounts use one without ever properly building the other. A Custom Audience is built from data you already have, customers, website visitors, engagement. A Lookalike takes one of those and finds new people who resemble it. Here's how to set up both correctly, from source data to a working Lookalike campaign.

Custom Audience Sources

Step-by-Step: Building a Customer List Custom Audience

1. Export your customer data as a CSV. Include email and phone number columns if you have both, more identifiers per person improves Meta's match rate against its own user base.
2. Go to Audiences in Events Manager and select "Custom Audience." Choose "Customer list" as the source.
3. Upload the file and map the columns. Meta asks you to confirm which column is email, phone, and so on, get this mapping right or the match rate drops significantly.
4. Wait for processing. Matching typically completes within a few hours, Meta shows you the estimated audience size once done, expect a match rate meaningfully below 100%, this is normal.
5. Name it specifically. "Purchasers - Last 12 Months" tells you far more six months from now than "Customer List 1."

Step-by-Step: Building a Lookalike From That Audience

1. From Audiences, select "Lookalike Audience." Choose your Custom Audience as the source, the quality of the Lookalike depends entirely on the quality of this source.
2. Choose the source country. This determines the pool Meta searches for similar users, match it to where you actually want to advertise, not just where the source audience happens to be.
3. Set the audience size percentage. Start at 1%, the closest match to your source audience and typically the highest-performing starting point, only widen to 2-5% once 1% is exhausted or underperforming.
4. Let it build. Lookalikes typically take under 24 hours to populate and refresh periodically as Meta's data updates.

Common Mistakes

• Building a Lookalike from too small a source audience. Under 100 people gives Meta too little signal to find a meaningful pattern, the resulting Lookalike often performs no better than a broad interest-based audience.
• Using "all website visitors" as the only website Custom Audience. This mixes someone who bounced in two seconds with someone who nearly converted, segment by page or time-on-site instead of treating all visitors as one audience.
• Never excluding existing customers from acquisition campaigns. A Lookalike or interest campaign that keeps showing ads to people who already bought wastes budget reaching people who don't need to be convinced again, exclude your purchaser Custom Audience from cold campaigns.

Frequently Asked Questions

How big should my source audience be for a good Lookalike?

Meta recommends at least 100 people, but quality matters more than the minimum, a highly specific, high-value source of 500 genuine purchasers typically outperforms a vague 5,000-person source.

Should I use a 1% or broader Lookalike?

Start at 1% for the closest match and best typical performance, only widen once that audience is saturated or too small to spend your budget efficiently against.

Do Custom Audiences update automatically?

Yes, both website and engagement-based Custom Audiences refresh automatically as new visitors or engagers qualify, uploaded customer lists need to be manually refreshed or connected through an integration for automatic updates.

The gap between an account running only broad interest targeting and one with a properly built Custom Audience and Lookalike structure is usually one of the largest single improvements available without increasing budget at all. If you want your audience structure built around your actual customer data rather than starting from a blank interest list, our Meta Ads team can set it up correctly.