Broad vs Interest vs Lookalike Targeting on Meta: What Works Now

For most e-commerce brands in 2026, broad targeting, usually through Advantage+ audience, is the best starting point, because Meta's delivery system finds buyers from purchase data and your creative does the real targeting. Interest and lookalike audiences still win in a few cases: new pixels with little data, narrow niches, and when you genuinely must limit who sees an ad. Test them against broad with identical creatives and budgets.

What is the difference between broad, interest and lookalike targeting?

Each tells Meta who to start looking for, but they give it very different amounts of room.

• Broad targeting sets only the basics, such as location, age and perhaps language, and leaves the rest to Meta. The system uses its own signals, such as who has bought from similar ads, to decide who sees yours.
• Interest targeting, part of what Meta calls detailed targeting, adds interests, behaviours and demographics, such as people interested in yoga or recently engaged shoppers.
• Lookalike audiences start from a source, such as a customer list or website purchasers, and find new people who share their characteristics. You choose a percentage of the population: smaller percentages match the source more closely, larger ones are broader.

How did Advantage+ audience change targeting?

Advantage+ audience is now the default audience setting for Sales campaigns in most accounts, and it changes what your targeting choices mean.

Meta's help pages split audience settings into two kinds:

• Audience controls are hard limits: locations, minimum age, custom audiences to exclude and languages. Meta won't show ads outside them.
• Audience suggestions are hints: age ranges, gender, detailed targeting and custom audiences to include, which covers lookalikes. Meta starts with people who match, then goes beyond them when it predicts better results.

So if you add an interest or a lookalike under Advantage+ audience, you are not targeting that group. You are telling Meta where to start. Meta's own example: if you suggest women, your ads could still reach men if the system finds them likely to respond.

You can still make interests and lookalikes hard limits. Choose the option to further limit the reach of your ads and untick "Use as a suggestion". That switches the audience section to Advantage+ off, and the campaign becomes a manual one.

There are also two older expansion settings worth knowing. Advantage+ detailed targeting lets delivery go beyond your interests, and Advantage+ lookalike lets delivery go beyond your chosen lookalike percentage. Meta says Advantage+ lookalike is switched on automatically for new ad sets that use lookalikes built from website, app or customer-list audiences.

When does broad targeting win?

Broad wins when Meta has enough signal to find buyers on its own. That usually means:

• Working purchase tracking. The Meta Pixel and Conversions API send purchases with value, so the system can learn who buys.
• Some purchase history. A pixel that has seen a few hundred purchases gives delivery much more to work with than a new one.
• Mass-market or wide-appeal products, such as apparel, skincare, home goods or coffee, where buyers are spread across many interest groups.
• Strong, specific creative. An ad about knee pain for runners pulls in runners with knee pain, whatever the targeting. This is why buyers say creative is the new targeting.
• Enough budget for a large audience. Broad audiences give the system room to keep finding new people as you scale, which slows audience saturation. Rising frequency is the early warning sign.

Broad also simplifies structure. Instead of five interest ad sets each starving for budget, one broad ad set collects all the learning in one place, which helps it exit the learning phase.

When do interest audiences still help?

Interests have lost most of their power as hard limits, but they are still useful in four situations.

1. New pixels. With no purchase data, a sensible interest suggestion gives delivery a starting point. Once purchases flow, test removing it.
2. Narrow niches. If you sell spare parts for a particular boat engine or kit for a minority sport, a broad audience may spend a lot finding the few relevant people. A focused interest can shorten that search.
3. Creative testing on small budgets. Some buyers add a relevant interest suggestion to tests so early spend lands closer to the likely buyer, then scale winners broad.
4. Messaging tests. Two ad sets with different interest suggestions and matching creative, such as gym-goers versus new parents, can show which segment responds to which angle.

Interest targeting is not a fix for weak creative. If an ad fails broad, it rarely succeeds narrow.

When do lookalike audiences still win?

Lookalikes still help when you have a strong source and want to point delivery at a specific kind of buyer.

What makes a good lookalike source?

Meta's help page says a source needs at least 100 people and generally recommends 1,000 to 5,000. Quality matters more than size. Meta notes you may get better results from your best customers than from all your customers.

Two recent changes are worth knowing:

• No location at creation. At the time of writing, Meta no longer asks for a location when you create a lookalike. It uses the ad set's location instead, and shows the estimated size there.
• Lookalikes as suggestions. Under Advantage+ audience, a lookalike is a suggestion, and Advantage+ lookalike may expand delivery beyond the percentage you chose.

Where lookalikes win

• Mature accounts with a clear best-customer list, such as subscribers or repeat buyers, where pointing delivery at that profile improves order value.
• Brands entering a new country with a customer list from an existing market.
• Accounts where broad drifts to low-value buyers, for example lots of discount-only first orders.

The lookalike audiences guide covers building and refreshing them step by step.

Broad vs interest vs lookalike: side by side

How do you test broad vs interest vs lookalike fairly?

Audience tests only mean something if everything except the audience stays the same.

1. Pick one question. For example: does a purchaser lookalike suggestion beat a plain broad audience for our hero product?
2. Use Meta's A/B test tool or separate ad sets on equal budgets. The built-in A/B test (Experiments) splits the audience so the same person doesn't see both versions, which gives the cleanest result. Separate ad sets on ad set budgets are simpler but can overlap. The split testing entry explains the difference.
3. Use identical creatives in every ad set: same ads, same copy, same landing page.
4. Exclude recent purchasers from every version, or none, so retargeting doesn't flatter one side. See exclusion audiences.
5. Fund each side properly. Aim for at least 20 to 30 purchases per side, which on most budgets means at least seven days. Use the CPA calculator to estimate the spend.
6. Judge on cost per purchase and new-customer share, not on CPM or click-through. Cheap clicks from a broad audience can still be expensive purchases, and vice versa.
7. Decide in advance what counts as a win, for example 15% lower cost per purchase, and keep the winner as your default until the next test.

How do you set up a broad ad set?

A broad ad set takes a couple of minutes, but a few settings decide whether it is genuinely broad or quietly narrow.

1. Locations: the countries or regions you actually ship to, set as an audience control. If your business can never advertise in some places, set that once in Advertising settings so every campaign inherits it.
2. Age: leave the full adult range, or set a minimum age as a control only if your product requires one. Meta lets you raise the minimum age up to 25 without turning Advantage+ off.
3. Gender and interests: leave them empty, or add them as suggestions if you have a real reason, such as a product made only for women.
4. Custom audiences: add recent purchasers as an exclusion if you want the ad set to focus on new customers.
5. Languages: leave empty unless you are running ads in a language many people in the location don't read.
6. Placements: keep Advantage+ placements on, with creatives in 4:5 and 9:16 so every placement looks native.

Then check the estimated audience size. For most countries it should be in the millions. If it looks small, an exclusion or an account-level control is probably narrowing it more than you meant.

What audience structure should an e-commerce brand use?

A setup that suits most growing stores:

• Scaling: one Advantage+ sales campaign or one broad Sales campaign, with existing customers defined so you can see how much spend reaches them.
• Testing: an ad set budget campaign testing new creative on a broad audience, or with one relevant suggestion on a new pixel.
• Occasional audience tests: a lookalike from your top customers against broad, run as an A/B test every quarter or so.
• Retargeting: a small separate campaign once your visitor pools are large, with its own creative. See Facebook retargeting ads.

What are the most common targeting mistakes?

1. Stacking many small interest ad sets. Each gets too little budget to learn. Consolidate into one broad ad set.
2. Judging audiences with different creatives. You end up testing the ads, not the audience.
3. Expecting interests to rescue a weak ad. If it fails broad, change the ad.
4. Building lookalikes from low-quality sources. All visitors and engagers make weak seeds.
5. Forgetting existing customers. Broad and Advantage+ audience include past buyers unless you exclude them, which can flatter results.
6. Calling a winner too early. A two-day audience test on a handful of purchases tells you almost nothing.

Choose the room, then fill it with creative

The audience question has shrunk. In most current campaigns, Meta treats your interests and lookalikes as starting points, and the strongest lever you control is what the ad says and shows. Start broad when you have purchase data, use interests and lookalikes as suggestions where you have a real reason, test them fairly with identical creatives and spend your energy on the creative testing framework that actually moves results.