Features
Ad Scheduling Impression Caps Super Title Exclusions HubSpot Attribution
Solutions
ABM Teams Demand Gen CMOs & VPs SaaS Startups Agencies HubSpot Users
Industries
HR Tech Cybersecurity Fintech Healthcare IT DevTools Legal Tech EdTech & L&D Martech
Resources
Blogs Budget Calculator Waste Calculator ROAS Guide Audit Checklist Attribution Guide LinkedIn vs Google Retargeting Guide Benchmarks 2026
Guide
Recession Budget Privacy Tracking Ads Changes Ads Ai Q4 Strategy
Comparisons
vs Metadata vs Dreamdata vs HockeyStack vs Bizible vs Manual Excel
Campaign Types
Retargeting Thought Leadership Lead Gen Forms Video Ads Document Ads Conversation Ads
Fix Problems
Fix High CPL Fix Low CTR Not Converting? Scale LinkedIn Ads Fix Ad Fatigue Small Audience?
Start Free Trial

Quick Summary

Summarize this article instantly with your preferred AI model.

LinkedIn Lookalike Audiences: What Happened and What to Use Instead


LinkedIn Lookalike Audiences: What Happened and What to Use Instead

LinkedIn Lookalike Audiences: What Happened and What to Use Instead

LinkedIn lookalike audiences no longer exist. LinkedIn discontinued the feature on February 29, 2024, so you can no longer create or edit a lookalike audience, and any you had built stopped refreshing and were eventually archived. If you came here looking to set one up, the short answer is that you cannot, but you can get the same effect, and arguably a better one, with the tools that replaced it: Predictive Audiences, Audience Expansion, and Matched Audiences. This guide explains what happened, what each replacement does, and how to rebuild lookalike-style targeting today. It also covers the part that matters more than which feature you use: the quality of the seed you feed it, because a lookalike or a predictive audience is only ever as good as the list it is built from.

Key takeaways

  • LinkedIn retired lookalike audiences on February 29, 2024; you can no longer create or edit them, and existing ones were archived after a short grace period.
  • The direct replacement is Predictive Audiences, which uses LinkedIn’s AI to build an audience likely to convert from a seed of 300 to 300,000 people.
  • Audience Expansion is a toggle that extends delivery to people similar to your target audience, and Matched Audiences let you target your own lists and website visitors.
  • Whichever tool you use, the audience is only as good as its seed: feed it your best customers or qualified leads, not every form fill you have ever collected.
  • Judge the resulting audience on qualified pipeline, not on reach or clicks, because efficient targeting of the wrong people is still waste.

Are LinkedIn lookalike audiences gone?

Yes. LinkedIn discontinued lookalike audiences, a feature it had launched in 2019, effective February 29, 2024. As of that date, advertisers could no longer create new lookalike audiences or edit existing ones, and the existing audiences stopped refreshing with new data and became static. Campaigns that were already running kept delivering to the frozen audience for a short time, and LinkedIn gave a roughly 30-day grace period before archiving the old lookalikes entirely. The change also flowed through to the API, so third-party platforms lost the ability to create lookalikes too.

LinkedIn did not give a detailed public reason for the retirement. The most likely explanation is a shift in approach: away from lookalikes, which find people who resemble your past audience, and toward AI-driven prediction, where LinkedIn’s own models decide who is likely to convert in the future. It is worth knowing that LinkedIn is now the outlier here, since most other major platforms, including Meta, Google, Microsoft Advertising, and TikTok, still offer lookalike targeting. On LinkedIn, though, the feature is gone for good, so the only question that matters is what to use instead.

What were LinkedIn lookalike audiences?

Lookalike audiences took a source audience you already had, such as a list of contacts or your website visitors, and expanded it into a larger audience of LinkedIn members who resembled that source on attributes like company, industry, and seniority. The promise was simple: show LinkedIn your good customers, and it would go find more people like them. It was a popular way to scale beyond a narrow list without hand-building a new target audience from scratch.

The weakness of the old feature, and the thing most people never examined, was that it resembled your seed on surface traits, not on whether those people would actually buy. If your seed list was a mix of great customers and random form fills, the lookalike faithfully found more of both. That limitation carries straight over to the tools that replaced it, which is why the seed matters more than the feature.

What replaced LinkedIn lookalike audiences?

LinkedIn pointed advertisers to three tools, which together do everything lookalikes did and more.

OptionHow it worksSeed or list size
Predictive AudiencesLinkedIn’s AI builds an audience likely to convert from a seed of your data: conversions, Lead Gen Form data, Insight Tag data, or a contact list300 to 300,000 people
Audience ExpansionA setting that extends delivery to members who share characteristics with your target audienceNone stated
Matched AudiencesTarget your own data: contact list uploads, website retargeting via the Insight Tag, company lists, or CRM integrationsA match rate under 5% means fewer than 300 matched

Predictive Audiences is the direct replacement and the closest thing to the old lookalike, except that it uses machine learning to predict conversion likelihood rather than just matching surface traits. You can hold up to 30 of them per ad account, they cannot be shared between accounts, and a campaign using one cannot also use audience expansion. Audience Expansion is the lightweight option: a toggle you turn on over an existing audience to let LinkedIn broaden delivery to similar members. Matched Audiences is the foundation under both, since your uploaded lists, website visitors, and CRM segments are the seeds you feed the other two. One point of confusion worth clearing up: a Matched Audience is built from your own data, while a Saved Audience is just a reusable set of targeting criteria like job title and industry, so the two are not the same thing and neither one is a lookalike.

How to recreate a lookalike-style audience now

To rebuild the old lookalike effect, use Predictive Audiences. In Campaign Manager, go to your audiences, create a new Predictive Audience, and choose a data source: a conversion event, Lead Gen Form submissions, Insight Tag data, or an uploaded contact list, as long as it has at least 300 people. LinkedIn’s model then builds an audience of members it judges likely to convert based on that seed, which you select as your targeting when you build the campaign. That is the modern equivalent of showing LinkedIn your good customers and asking for more like them.

A few concrete examples of the pattern:

  • Seed a Predictive Audience with your closed-won customer list, so LinkedIn looks for more members who resemble people who actually bought, not just people who filled a form.
  • Turn on Audience Expansion over a tightly defined audience, for instance a banking-industry audience, to reach adjacent finance roles you did not name explicitly.
  • Build a Matched Audience of recent website visitors for retargeting, then save it so you can reuse it across campaigns without re-uploading.

Whichever route you take, start small, test the new audience against your existing targeting before scaling, and keep the underlying data fresh. And remember you cannot combine Predictive Audiences with Audience Expansion in the same campaign, so pick one approach per campaign.

The seed decides the audience

Here is the part that matters more than any feature. Both the old lookalikes and the new Predictive Audiences are machines for finding more people like the ones you show them, which means the single biggest lever on the result is the quality of the seed, not the cleverness of the targeting. Feed the AI a seed of your best closed-won customers, and it looks for more buyers. Feed it a seed of every lead you ever captured, junk included, and it efficiently finds more junk. The targeting feature cannot tell the difference, because it optimizes toward resembling the seed, whatever the seed is.

So the work that actually improves a predictive or expanded audience is upstream of LinkedIn. Build your seed from the people you actually want more of: closed-won customers, or at least sales-accepted leads, not raw form fills. The tighter and higher-quality the seed, the better the audience, and a seed of 300 genuinely good customers will usually outperform a seed of 3,000 mixed leads. This is the same principle that governs everything in LinkedIn Ads: the platform will efficiently deliver whatever you optimize toward, so point it at quality. And judge the resulting audience the only way that counts, by the qualified pipeline it produces, not by how many people it reached or how cheap the clicks were. An audience that is cheap to reach and never converts is not a good audience.

Predictive Audiences vs the old lookalike audiences

If you used lookalikes before, here is how the replacement compares.

AspectLookalike Audiences (retired)Predictive Audiences (now)
StatusDiscontinued February 29, 2024The current replacement
How it is builtExpanded a seed to a broad audience with similar traitsAI scores members and builds an audience likely to convert
What it matches onSurface attributes like company and seniorityPredicted likelihood to convert, learned from your data
Seed sourceA matched audience seedConversions, Lead Gen Form data, Insight Tag data, or a contact list
Seed sizeAround 300300 to 300,000

The headline difference is that Predictive Audiences try to predict who will convert, not just who looks similar, which in principle is an upgrade. But the upgrade only pays off if your seed reflects conversions worth having. Point either tool at a weak seed and you get a weak audience, so the move from lookalikes to predictive audiences changes the engine, not the rule that the seed decides the outcome.

Common mistakes with LinkedIn audience targeting

The biggest mistake now is still building audiences from a low-quality seed, feeding Predictive Audiences every lead you have instead of your best customers, then wondering why the audience underperforms. The second is judging the audience by reach or cost per click rather than by qualified pipeline, so an efficient but unqualified audience looks like a win. The third is skipping the test, scaling a new predictive or expanded audience before proving it beats your existing targeting. The fourth is letting the seed go stale, since an audience built on last year’s customer list drifts away from who you sell to now. And the fifth is trying to combine Predictive Audiences with Audience Expansion, which LinkedIn does not allow in the same campaign.

Where this leaves you

So, LinkedIn lookalike audiences are gone, retired on February 29, 2024, and the replacement is better in principle: Predictive Audiences use AI to find people likely to convert, with Audience Expansion and Matched Audiences filling in around them. But the change does not let you off the hook for the thing that always mattered most. Whichever tool you use, build the seed from the customers you actually want more of, test before you scale, keep the data fresh, and measure the audience on qualified pipeline rather than reach. The feature changed; the rule that your audience is only as good as its seed did not. If you want your LinkedIn targeting built from the right seeds and measured on pipeline, you can book a demo.

Frequently Asked Questions

Q1. Are LinkedIn lookalike audiences still available?

No. LinkedIn discontinued lookalike audiences on February 29, 2024. Since then you cannot create new lookalike audiences or edit existing ones, and the lookalikes people had already built stopped refreshing and were archived after a short grace period. The change also applied through the API, so third-party tools can no longer create them either. If you want lookalike-style targeting, the replacement is Predictive Audiences, which uses LinkedIn’s AI to build an audience likely to convert from a seed of your data.

Q2. When did LinkedIn retire lookalike audiences?

LinkedIn retired lookalike audiences effective February 29, 2024, having first launched the feature in 2019. On that date, the ability to create and edit them was removed, existing lookalikes became static and stopped updating, and campaigns already using them kept delivering to the frozen audience for a short time. LinkedIn then archived the old audiences after roughly a 30-day grace period. LinkedIn did not publish a detailed reason, but the move fits its broader shift toward AI-driven targeting through Predictive Audiences.

Q3. What replaced LinkedIn lookalike audiences?

Three tools replaced them. Predictive Audiences, the direct replacement, uses LinkedIn’s AI to build an audience likely to convert from a seed of 300 to 300,000 people, drawn from conversions, Lead Gen Form data, Insight Tag data, or a contact list. Audience Expansion is a toggle that extends delivery to members similar to your target audience. Matched Audiences let you target your own contact lists, website visitors, company lists, and CRM segments, and they serve as the seeds for the other two. Together they do everything lookalikes did and add conversion prediction.

Q4. What is the difference between predictive and lookalike audiences on LinkedIn?

Lookalike audiences, now retired, expanded a seed into a larger audience of members with similar surface attributes like company and seniority. Predictive Audiences, the replacement, use machine learning to build an audience of members predicted to convert, based on patterns in your seed data rather than just resemblance. In principle predictive audiences are an upgrade, because they aim at likelihood to buy rather than similarity. In practice both depend entirely on the seed, so a weak seed produces a weak audience with either tool.

Q5. How do you create a lookalike-style audience on LinkedIn now?

Use Predictive Audiences. In Campaign Manager, create a new Predictive Audience and choose a data source with at least 300 people: a conversion event, Lead Gen Form submissions, Insight Tag data, or an uploaded contact list. LinkedIn’s model builds an audience it judges likely to convert from that seed, which you then select when building your campaign. For a lighter option, turn on Audience Expansion over a Matched Audience. Either way, build the seed from your best customers, and test the new audience against your existing targeting before scaling.

Q6. What is the minimum audience size for LinkedIn Predictive Audiences?

Predictive Audiences require a data source of at least 300 people, and accept seeds up to 300,000. That seed can be a conversion audience, Lead Gen Form data, Insight Tag data, or an uploaded contact list, and the quality of those people matters more than the quantity. You can hold up to 30 Predictive Audiences per ad account, they cannot be shared between accounts, and a campaign using a Predictive Audience cannot also use Audience Expansion. If you are uploading a list, note that a match rate under 5 percent means fewer than 300 members were matched.

Q7. What is Audience Expansion on LinkedIn?

Audience Expansion is a setting you turn on over an existing audience to let LinkedIn broaden delivery to members who share characteristics with your target, such as similar job titles, companies, or demographics. It is the lightweight way to reach beyond your exact audience without building a separate Predictive Audience. It works with Matched Audiences and attribute targeting, but it is not available for some ad formats, and it cannot be combined with Predictive Audiences in the same campaign. Use it to widen a tightly defined audience when you need more reach.

Q8. Do other platforms still have lookalike audiences?

Yes. LinkedIn is the outlier here. Most other major advertising platforms, including Meta (Facebook and Instagram), Google, Microsoft Advertising, and TikTok, still offer lookalike or similar-audience targeting. So if you run ads across channels, you can keep using lookalikes elsewhere; it is only on LinkedIn that the feature has been retired. On LinkedIn, the equivalent is now Predictive Audiences, which aims to improve on the old lookalike model by predicting who is likely to convert rather than just finding people who resemble your seed.