Yes, lookalike audiences still work on Meta ads, but they work differently from the way most guides describe. Meta now takes the audience location from the ad set instead of from the moment of creation, and it switches on Advantage+ lookalike for new ad sets that use a lookalike, which lets delivery go outside the audience when Meta expects a better result. Inside Advantage+ audience a lookalike becomes a suggestion unless you make it a hard limit, so the question to ask is not whether lookalikes work but what your source audience teaches Meta.
The rules Meta applies to a lookalike today
| Setting | What Meta documents |
|---|---|
| Source size | at least 100 people, with 1,000 to 5,000 recommended |
| Lookalikes per source | up to 500 |
| Audience size | 1% to 10% of the population of the ad set’s locations, with 1% as the closest match |
| Location | taken from the ad set, not chosen when you create the lookalike |
| Populating | up to 3 days, and ads can run in the meantime |
| People in the source | excluded from the lookalike |
| A saved audience as source | not allowed |
The 1% to 10% range comes from Meta’s Leads Center article, and the rest from its article on creating a lookalike, so both apply to a Leads campaign. A side effect of the location change is visible in Audience Manager: a new lookalike shows no estimated size until you use it in an ad set. Add several lookalikes to one ad set and Meta targets people in any of them, which is the usual way to cover a country without building one 3% audience.
The step by step version, with the source types you can choose from, is in how to create a lookalike audience in Meta ads.
Advantage+ lookalike and Advantage+ audience: a hint, not a fence
Two different behaviours sit under similar names, and agencies mix them up.
Advantage+ lookalike expands delivery. Meta turns it on automatically for new, duplicated and draft ad sets that use a lookalike built from an app, website or customer list custom audience, and it lets the ad set reach people outside the lookalike when that looks likely to improve performance. It is the stricter of the two to switch off, because it changes what the ad set can do rather than how the audience is labelled.
Advantage+ audience turns the lookalike into a suggestion. In Sales, App promotion and Leads campaigns, a lookalike sits under Custom audience inclusions in the Add suggestions panel, which is where Advantage+ audience reads it as a hint about who to look for. To make the lookalike a real boundary, open Further limit the reach of your ads, select it again and uncheck Use as a suggestion.
The limits follow from that. A 1% lookalike added as a suggestion does not hold delivery inside 1% of the population, so reading results per audience from such an ad set measures Meta’s choices as much as your list. And an exclusion of a lookalike is not available at all while Advantage+ audience is on.
The source audience decides whether the lookalike still works
Meta looks for people who resemble whoever is in the source, so the source is the whole strategy. Meta’s own advice is that a source built from your best customers can beat one that includes all of them, and lead generation has the same split: leads that qualified or closed against every form fill.
Example: you have 2,000 leads from the last year and 300 of them became customers. A lookalike of the 2,000 teaches Meta what a person who fills in a form looks like. A lookalike of the 300 teaches it what a customer looks like, and 300 still clears the 100-person minimum. The same file can produce both, so the choice is about which one you upload.
Which leads qualified is a decision your CRM makes, not Meta. The busiest route is to export the qualified segment as a customer list and build the lookalike from it. Meta’s Leads Center can also build custom and lookalike audiences from your leads once you have generated more than 20 on Meta, which shortens the path but does not tell you which leads were good. For the setup around the list, the lookalike audience entry covers the terms and the related settings.
A flagged source is the other failure mode. If the custom audience behind a lookalike is flagged for information Meta’s terms do not allow, the lookalike is flagged too, and the repair starts with the list rather than with the ad set.
How to tell whether a lookalike still works in your account
The only honest test is one where the audience is the single thing that changes.
- Pick one outcome you already measure, such as a booked appointment or a qualified lead, and use it for both arms.
- Build the source list from qualified leads, and write down how many people it holds.
- Create the ad set you would normally run, then duplicate it and change nothing but the audience: the same creative, the same budget, the same dates.
- In the arm with the lookalike, decide in advance whether it is a suggestion or a hard limit, and leave Advantage+ lookalike on or off for both arms consistently.
- Run long enough to compare a real number of outcomes, not one week of activity, and read lead quality from your own records rather than from Meta’s modelled count.
- Repeat with the other source, for example all leads against qualified leads, and keep the winner as your standard source.
Meta’s own material makes the same point about suggestions: if you want to know whether the lookalike itself works, make it a control in at least one arm of the test. Without that arm you are testing the delivery system.
Where lookalikes are limited or unavailable
Special ad categories are the first place to check. For housing, employment and financial products and services campaigns from US-based advertisers, or reaching the US, Canada or certain European countries, Meta lists lookalike audiences as limited or unavailable, and Advantage+ lookalike as unavailable. A mortgage or insurance advertiser in Europe can find that the audience type it planned to use is not on offer at all, before a single euro is spent.
The second limit is technical: a saved audience cannot be a source, so you need a custom audience built from a customer file, site activity or engagement rather than a saved targeting set. The third is size. A source under 100 people produces no lookalike, and a long list of dead form fills produces one that finds more of the same.
What to do instead when the lookalike is the wrong tool
If the source is too small, a broad ad set with good creative is a reasonable starting point, and campaign setup covers the structure choices around it. If the problem is lead quality rather than audience size, the other lever is sending qualified outcomes back to Meta, which is what the conversion leads setup does: it points delivery at people likely to become good leads instead of people likely to fill in a form.
Both routes still end in the same measurement question. DropCPL ties every lead to the ad, the creative and the landing page that produced it, and computes cost per lead from the leads your landing pages received, so the comparison between two audiences rests on the leads themselves rather than on modelled numbers.