Create a Lookalike Audience on Meta

Show your ads to new people who look like your best customers. The fastest way to scale beyond retargeting.

5–10 minutes

Verified by Hassan Rawther ·

Hassan Rawther is the Founder, Unlearn Academy. Written by the Unlearn Academy team.

0 of 4 doneContinue →

The short answer

To create a lookalike audience on Meta, open Audiences in Ads Manager, click Create audience → Lookalike audience, and choose a source with at least 100 people from one country, such as purchasers or leads. Pick the location and an audience size from 1% (most similar) to 10% (biggest), then click Create. It is usually ready within a few hours.

Before you start

By the end of this SOP

A lookalike audience marked “Ready” that you can add to any ad set.

Time
5–10 minutes
You’ll need
An ad account (SOP 010) and a source audience: pixel events (SOP 017), a customer list, or Page/Instagram engagers.
Who does this
The person responsible for paid advertising.
About this SOP: why it matters, where and when

Why it matters: Your customers already tell Meta who buys. A lookalike turns that into a new audience of similar people.

Where: In Meta Ads Manager → Audiences.

When: When you want to scale prospecting campaigns.

Stage 1

Pick a strong source

A lookalike is only as good as its source. Best to worst: buyers, leads, website visitors, engagers.

  1. 1

    Good sources:

    • Purchasers or leads from your pixel (last 30–180 days).
    • A customer list uploaded as a custom audience.
    • People who engaged with your Instagram account or Facebook Page.
  2. 2

    The source needs at least 100 people from a single country. More is better; 1,000+ gives much better results.

    Tip

    Not enough purchasers yet? Use a step that happens more often, like Initiate checkout or Add to cart.

  3. 3

    Before using pixel data, sanity-check it: do the event counts and values match what you see in your store or CRM?

Stage 2

Create the lookalike

  1. 1

    In Ads Manager, open the menu → Audiences. Check the right ad account is selected.

  2. 2

    Click “Create audience” → “Lookalike audience”.

  3. 3

    Select your source. If you send purchase values, Meta may offer a value-based source, which favours your highest-value customers.

  4. 4

    Choose the location where you want to find new people, e.g. India, or Kerala via a later ad set filter.

  5. 5

    Choose the audience size: 1% is the most similar, 10% is the largest. Start with 1–3%.

    Note

    You can’t edit the size later. Create another lookalike instead. Up to 500 per source.

  6. 6

    Click “Create audience”. Wait until Availability shows “Ready”, often within a few hours, up to 24.

Stage 3

Use it in an ad set

  1. 1

    In your prospecting ad set, under Audience, add the lookalike in Custom audiences.

  2. 2

    Exclude your existing customers so you don’t pay to reach people who already bought.

  3. 3

    Test it against a broad audience with Meta’s A/B test tool (Experiments), not by eye.

    • Run it for at least 14 days, or until each side has about 50 results.
    • Decide on cost per qualified lead or sale from the client’s sheet, not only Meta’s cost per result.

Agency playbook

Doing this for a paying client

What a senior checks before calling the job done. Numbers are starting rules; compare them with the account’s own history.

What good looks like

  • The source is your best customers, not everyone: enrolled students or repeat buyers, not all leads.
  • Customer-list sources are refreshed at least monthly (starting rule). Pixel-based sources update by themselves.
  • The test is judged on cost per qualified lead or sale, same dates and same budget, not on CTR (click-through rate).
  • The prospecting ad set excludes current customers and recent leads, so the lookalike finds genuinely new people.

Mistakes we see on real accounts

  • Building from all leads, junk included. Meta finds more people like the junk, and lead quality falls further.
  • Running 1%, 2% and 3% lookalikes in separate ad sets. They overlap heavily, compete with each other and none exits learning.
  • Uploading a customer list without consent. It risks breaking DPDP rules and Meta’s terms, and can get the account restricted.
  • Never refreshing the customer list. After six months the source no longer reflects who buys today.

Agency checklist before you call it done

  • The client has consent to use the customer list for ads, and their privacy policy says so (India’s DPDP Act, Digital Personal Data Protection Act).
  • List formatted: phones with +91, emails in lowercase, plus name and city for a better match.
  • Upload directly in Ads Manager, where Meta hashes the data. Never send raw customer lists over WhatsApp or personal email.
  • Source and lookalike named per convention, e.g. “CRM-enrolled-IN-180d” and “LA-1-IN-CRM-enrolled”.
  • Screenshot of the audience showing “Ready” and its size, saved in the client folder.
  • Change-log entry: source, date range, size, and which ad set uses it.
  • Local copies of the customer file deleted after upload, as per your agency data policy.

Handover message to the client

Hi [Client name], we built a lookalike audience from your [1,200] enrolled students (uploaded securely, raw file deleted). We’re testing it against broad targeting for 14 days at ₹[1,000]/day each. We’ll compare cost per enrolled student on [date].

Senior tip

Lookalikes built from website visitors often include people who already know you, so results look better than they are. Exclude recent visitors and leads, or run a lift test, before you call a lookalike a winner.

Practice & check

Now do it for real

Reading is not doing. Open your own account and complete this task before you move on.

Practice task (5 min)

Create a 1% lookalike from your Instagram engagers or a pixel event in your practice account. Screenshot it showing “Ready”.

Prove it

1. What is the minimum source size for a lookalike?
2. Which lookalike size is most similar to your source?

Two questions. Get both right and this SOP is done.

Common questions

How long does a lookalike audience take to be ready?

Usually a few hours, sometimes up to 24 hours. Availability shows “Ready” when you can use it.

Do lookalike audiences still work in 2026?

Yes, but with Advantage+ audience they act as a suggestion rather than a hard limit. They still help Meta find the right people faster, especially for new accounts.

What is a value-based lookalike?

A lookalike built from customers weighted by how much they spent, so Meta finds people similar to your highest-value buyers. It needs purchase values sent with your pixel events.

Done. That’s a real skill, not a certificate.

Next: SOP 075 — Plan Your Meta Ads Campaigns →

Want to do this on real client budgets, with a mentor checking your work? That’s Level 2 at Unlearn. See the Digital Marketing track