Lookalike audience

A lookalike audience is an ad audience an ad platform builds to match people similar to a list you give it, like your current customers or recent demo requests. Meta still calls it a Lookalike Audience, while LinkedIn replaced its version with Predictive audiences in 2024.
What is a lookalike audience?
A lookalike audience is a group of people an ad platform builds for you because they resemble a list you already have. You give the platform a source list, often called a seed, like your customers or recent demo requests. The platform looks at what those people have in common and finds new people who share those traits.
Let's say you sell expense management software. You upload 800 contacts from your current customers to an ad platform. It finds thousands of people with similar jobs, companies and online behavior, and shows them your ads. You never had to guess which job titles or industries to target.
That's the main appeal. A lookalike targets people based on who your buyers really are, which can be hard to describe with the targeting boxes in an ad manager.
Lookalike audiences on the big ad platforms in 2026
The name and the rules changed a lot over the last few years, so it helps to know what each platform offers today:
Here's what that looks like on each platform:
- Meta: Meta still offers Lookalike Audiences. You pick how close the match should be, and a 1% audience is the closest match to your source.
- LinkedIn: LinkedIn discontinued lookalike audiences on February 29, 2024. It now recommends Predictive audiences, which you build from contact lists, company lists, conversions, Lead Gen Forms or retargeting audiences. They refresh daily and can take up to five days to reach full size.
- Google Ads: Similar audiences ended on May 1, 2023. Google now offers Lookalike segments in Demand Gen campaigns, and during 2026 it's moving them to a suggestion mode, where your seed list guides Google's targeting instead of limiting it. Advertisers can opt out.
Why lookalike audiences matter for B2B SaaS
B2B audiences are small and expensive. If you sell to QA leads at software companies with 200 to 2,000 employees, job title targeting alone gets narrow fast. You end up showing ads to the same few thousand people over and over.
A lookalike audience lets you reach people outside your lists who still look like your buyers. I think it's most useful when your best customers share traits that targeting fields can't capture, like a certain tech stack or a way of working. It also pairs well with retargeting, since lookalikes bring new people in and retargeting follows up with the ones who visited.
How to build a good lookalike audience
The seed decides almost everything. A great seed with average settings will beat a weak seed with perfect settings. Here's my process:
- Pick your best seedStart with customers or contacts from closed-won deals. The platform copies whatever you give it, including the weak leads.
- Make it big enoughMeet the platform's minimum, then aim higher where you can. Adding recent demo requests to your customer list is a common way to grow it.
- Keep it freshSync the list from your CRM, or re-upload it every month, so new customers keep shaping the audience.
- Exclude who you already knowRemove current customers and open deals from your targeting so you don't pay to reach people you already talk to.
- Test it against your other targetingRun the lookalike next to job title or company list targeting and compare the cost per qualified lead.
On LinkedIn, Audience Expansion is turned off for ad sets that use Predictive audiences, according to LinkedIn's help page. So the predictive audience does the expanding, and you don't need to switch anything else on.
A lookalike audience example
Here's a hypothetical example. Let's say you sell a test automation tool and run ads on LinkedIn and Meta. You export 1,200 contacts from closed-won deals in HubSpot. On LinkedIn, you upload them as a contact list and build a Predictive audience from it. On Meta, you build a 1% Lookalike Audience from the same list.
After a month, you compare the two by cost per SQL and by how many leads match your ICP. LinkedIn might cost more per click and still win on qualified pipeline, because its data on job titles and companies is so strong for B2B. Meta might bring cheaper leads from smaller companies. Either way, you decide based on pipeline, and click price is only part of the picture.
How to measure a lookalike audience
Clicks and impressions are easy to get from a lookalike. The question is whether the people who click look like your customers. Here's what I track:
- Cost per qualified lead: Divide spend by leads that match your ICP, next to your overall CPL. Cheap leads from the wrong companies are expensive in the end.
- Lead to SQL rate: Compare it with your other campaigns. A lookalike that fills the funnel with leads sales rejects isn't working.
- ICP match rate: Check what share of lookalike leads fit your company size and job title rules.
- Pipeline and revenue: Tie lookalike campaigns to opportunities in your CRM with UTM parameters, so you can see revenue by audience.
Common lookalike audience mistakes
- Seeding with every lead: A list of all form fills includes students, job seekers and competitors. The platform will find more of them.
- Using an old list: Customers from three years ago may not match your ICP today. Refresh the seed when your product or market changes.
- Following old advice: Many guides still explain LinkedIn lookalikes or Google similar audiences. Check what each platform offers today before you plan a campaign.
- Judging on click price alone: The cheapest clicks often come from the widest audience. Judge each audience on pipeline.
- Forgetting exclusions: Without excluding current customers, part of your budget goes to people who already pay you.
Here's a question to ask before your next campaign. If the platform copied your seed list perfectly, would you want a thousand more of those people?
Read more
FAQ
Does LinkedIn still have lookalike audiences?
No. LinkedIn discontinued lookalike audiences on February 29, 2024. It recommends Predictive audiences for contact lists, company lists, conversions, Lead Gen Forms and retargeting audiences, and Audience Expansion for other targeting.
Does Google Ads have lookalike audiences?
Google Ads stopped generating similar audiences on May 1, 2023. Today it offers Lookalike segments in Demand Gen campaigns, and during 2026 Google is moving them to a suggestion mode, where the seed list guides targeting instead of limiting it.
How big does a lookalike source audience need to be?
Meta's developer documentation asks for at least 100 people in the source audience. LinkedIn's Predictive audiences need more than 300 members, or 300 to 300,000 rows for an uploaded list. Bigger, higher-quality seeds usually give better results.
What is the best seed list for a B2B lookalike audience?
Customers or contacts from closed-won deals are the strongest seed, because the platform copies whatever you give it. A list of every lead includes students, job seekers, competitors and poor-fit companies, and the lookalike will find more of them.



