LinkedIn retired lookalike audiences on February 29, 2024. The working alternatives to LinkedIn lookalike audiences are predictive audiences, Audience Expansion, and a lookalike contact list you build and own yourself.
I’m going to be honest with you. I watched this change bite a real campaign. In March 2024, a fintech team I advised was still paying to reach a lookalike segment that had quietly stopped updating. Nobody noticed for six weeks.
So let’s sort this out properly. What actually happened, what LinkedIn wants you to use now, and the one alternative most marketers never consider.
📌 TL;DR: LinkedIn discontinued lookalike audiences on February 29, 2024, and existing ones went static. Inside LinkedIn Ads, the replacements are predictive audiences (seeded by your lists, modeled by AI) and Audience Expansion. Outside the platform, you can build a person-level lookalike contact list you own, work it through outbound, and upload it as a matched audience whenever you do run ads.
What Happened to LinkedIn Lookalike Audiences?
LinkedIn discontinued lookalike audiences on February 29, 2024. No new ones can be created, and existing ones can’t be edited.
And the part that stung advertisers most sits right in LinkedIn’s own help center announcement:
“Existing lookalike audience data will no longer refresh and a lookalike audience will become a static lookalike audience.”
LinkedIn Marketing Solutions Help Center
Static means frozen. Your ad sets kept delivering, but to a snapshot of who looked like your customers back in early 2024. People change jobs, companies pivot, and a frozen audience drifts further from reality every month. LinkedIn also archived audiences that sat unused for 30 days and shut down the Lookalike API the same day.
Why kill a popular feature? LinkedIn framed it as a move toward AI-driven targeting, as Social Media Today reported when the change was announced. And LinkedIn wasn’t alone. Google had already replaced similar audiences with Lookalike segments and optimized targeting in its own ecosystem. The whole industry moved from “copy this seed” to “let the model decide.”
What should you do first? Open Campaign Manager and check whether any saved audience in your account still traces back to a pre-2024 lookalike. If one does, treat it as expired inventory. It won’t throw an error. It will just quietly underperform, which is worse.
That’s the history and the health check. Now let’s look at what LinkedIn handed you instead.
What Is a Predictive Audience on LinkedIn?
A predictive audience is an audience LinkedIn’s AI builds from a data source you upload. You hand it a seed (a contact list, a company list, conversion data, Lead Gen Form fills, or retargeting activity), and the system finds members it predicts will act like the people in that source.
The mechanics matter, so here they are in plain words. According to LinkedIn’s predictive audiences documentation, your data source needs at least 300 members. A contact list source runs from 300 up to 300,000 rows and gets uploaded as a CSV in Campaign Manager. Each ad account can hold up to 30 predictive audiences, and they can’t be shared across accounts.
Because the model learns from your source, the source is everything. The closer that list sits to your ideal customer profile, the better the raw material the AI has to work with. Feed it your 300 closest wins, not 300 random rows.
💡 Pro Tip: Don't launch a predictive audience at the bare 300-row floor. Give the model room. A few hundred extra high-fit rows produce a noticeably tighter audience than the legal minimum.
Predictive Audiences vs Lookalike Audiences: What Actually Changed
On paper, the swap looks simple: one similarity engine out, another in. But the two work differently in ways that change how you plan campaigns, so here’s the honest side-by-side.
| Lookalike audiences | Predictive audiences | |
|---|---|---|
| Status today | Discontinued Feb 29, 2024; existing ones frozen | Live, and LinkedIn’s named replacement |
| Source data | One matched audience you picked | Contact lists, company lists, conversions, Lead Gen Forms, retargeting |
| Refresh behavior | No longer refreshes (static) | Modeled continuously while campaigns run |
| Your control | None anymore; can’t create or edit | You control the source; the output stays algorithmic |
| Size rules | Not applicable now | 300 to 300,000 rows per contact list, 30 audiences per account |
See the trade? You give up manual audience surgery and get continuous modeling in return. The system behaves a lot like lead scoring turned outward: it studies your source and scores strangers by resemblance.
Does it work in practice? Early field reports are encouraging when the seed is strong. The team at Workshop Digital reported match rates around 75% on contact lists of 40,000 to 75,000 rows, with cost per lead dropping meaningfully after the switch. Their lists carried names, emails, job titles, companies, and countries. Thin lists don’t get those numbers.
What Are Your Alternatives Inside LinkedIn Ads?
Inside LinkedIn Ads, the alternatives are predictive audiences, Audience Expansion, and matched audiences built from your own lists. Each covers a different slice of what lookalikes used to do.
Audience Expansion
Audience Expansion widens an existing campaign audience by adding members whose professional attributes resemble your targeting. Skills, titles, interests. You target “Online Advertising” and it pulls in close cousins of that skill.
It’s the lowest-effort option, and that’s both the pitch and the problem. For broad awareness plays, fine. But for tight account-based work, expansion loosens your grip on exactly who sees the ad, which defeats the point of picking accounts in the first place. Most ABM practitioners I know keep it switched off and would rather control the list themselves.
Matched Audiences: Upload a List You Trust
The steadiest ads-side route is the least magical one: Matched Audiences, where you upload your own contact or company list and LinkedIn matches the rows to member profiles. No modeling, no guessing. Just the people you chose.
Here’s the catch, though. Match rates live and die on list quality. Stale titles, missing fields, and half-guessed emails mean LinkedIn can’t find the person, so your real audience shrinks. That’s why contact data enrichment before upload is the least glamorous, highest-return step in this whole playbook: complete records with clean names, companies, and firmographic data simply match better.
To be clear, this isn’t an ads tutorial. LinkedIn’s docs own the click-path. What I want you to take away is the pattern: every ads-side alternative starts with a list, and the list is the part you control. Which raises a bigger question. If the list is the valuable part, why rent the audience at all?
The Seller-Side Alternative: Build the Lookalike List Yourself
Every option above shares one weakness: the audience lives inside an ad platform. You can’t call it, email it, or export it. When LinkedIn changes the rules again (and February 2024 proved it will), your targeting evaporates with it.
A lookalike contact list flips that. Instead of asking a platform to find similar people and keep them behind glass, you generate the similar people as actual named contacts in a spreadsheet you own. Full disclosure: I work at CUFinder and we build exactly this, so weigh my enthusiasm accordingly. But the workflow is simple to judge on its merits.
With the Contact Lookalike Finder, you enter one person by work email or LinkedIn URL. The AI builds a 360° profile from that person’s posts, reactions, and activity, plus their company’s, and returns the 25 most similar contacts. It models the person, not a profile sidebar. For a whole seed list, you upload an Excel or CSV file with an email or LinkedIn URL column and run it in bulk. Results export to Excel, save as a list, or push straight into HubSpot, Salesforce, or Zoho. Credits are only charged on successful matches, so a “Not Found” row costs nothing. Developers can hit the same engine through the API with a single POST to the v2/clf endpoint.
The rhythm looks like this:
→ 1 closed-won champion → 25 similar contacts → qualify → outreach list this week.
🧠 Remember: Credits only burn on successful matches. Test with a handful of seed contacts first, check the quality of the 25 that come back, and only then run the full list.
Who should the seed be? Your best customer’s champion, the person who drove the deal, someone who matches your strongest buyer persona. One quick boundary note: this clones people. If you want to clone entire companies instead, that’s an account-level job, and the way to build a lookalike account list is a different workflow with different inputs.
Now the honest limits. A lookalike list is a starting point, not a pipeline guarantee, and every name still deserves qualification before outreach. And because each seed returns 25 contacts, you scale by adding more good seeds, not by squeezing one seed harder. Seed quality rules here too: a mediocre seed produces 25 mediocre matches, cheerfully and instantly. The upside is durability: the same list works in your sequences today and uploads as a matched audience the day you turn ads back on. One asset, two channels, zero landlords.
Which Alternative Should You Pick?
Running ads? Start with predictive audiences. Doing outbound or ABM? Build a lookalike list you own. Most teams end up needing both, so here’s the split I recommend:
- You spend real budget on LinkedIn Ads: seed a predictive audience with your cleanest 1,000+ contact rows and let it model.
- You run tight ABM on named accounts: skip expansion, upload matched audience lists, and keep control of every row.
- You live in outbound: make lookalike prospecting your default motion and let sellers work the 25-per-seed lists directly.
- You’re translating an ads habit into pipeline strategy: read up on how lookalike audiences in B2B actually behave before committing budget either way.
That’s the decision. Now let’s make sure you don’t repeat the mistakes I keep seeing.
Mistakes to Avoid When Replacing Lookalike Audiences
1. Leaving a static lookalike running unchecked. This is the one I lived through. That fintech team in March 2024? Their lookalike segment froze with the February cutoff, and for six quiet weeks the ads kept spending against an audience that no longer updated. Cost per lead crept up roughly 40% before anyone asked why. We rebuilt with a predictive audience seeded from a cleaned 2,400-row contact list, gave the SDR team an owned lookalike list for outbound, and I now keep a quarterly calendar reminder to audit every saved audience. Boring habit. Saves thousands.
2. Feeding the minimum and expecting magic. A 300-row source scraped together from old webinar signups is technically valid and practically useless. The model amplifies whatever you feed it, junk included.
3. Leaving Audience Expansion on for ABM. If you picked 80 target accounts on purpose, don’t let an algorithm quietly add strangers to the room.
4. Renting forever and owning nothing. If every audience you have lives inside Campaign Manager, your targeting has a landlord. Keep an exportable, CRM-synced version of your best-fit contacts so no platform change can zero you out again.
5. Judging any alternative in week one. Predictive audiences need processing and learning time, and outbound lists need a full sequence cycle. Give each test a fair window before you call it.
FAQ: Alternatives to LinkedIn Lookalike Audiences
What is a predictive audience on LinkedIn?
A predictive audience is an AI-modeled audience built from a data source you provide. LinkedIn studies your contact list, company list, conversions, or Lead Gen Form data, then targets members it predicts will behave like the people in that source. Contact list sources need 300 to 300,000 rows.
What is the difference between a custom audience and a lookalike audience?
A custom audience targets people you already know; a lookalike targets strangers who resemble them. Custom means your uploaded list, matched to real accounts. Lookalike means the platform models new people from that list. One is retention and reach-back, the other is net-new discovery.
When should I use a lookalike audience?
Use lookalike-style targeting when you have a proven customer set and need net-new reach beyond it. On platforms that still offer lookalikes, that means a strong seed of real customers. On LinkedIn, the same job now belongs to predictive audiences or to a lookalike contact list you build yourself.
Do lookalike audiences still work?
Yes on Meta, no on LinkedIn, and it depends on your seed everywhere. Meta still supports lookalike audiences, while LinkedIn replaced them with predictive audiences in 2024. Results track seed quality more than platform choice, which is why an owned, well-qualified contact list stays useful no matter what platforms change next.
What is the 3/2/1 rule on LinkedIn?
The 3/2/1 rule is an organic posting rhythm, not an ads targeting rule. Popularized by LinkedIn creators, it suggests roughly three industry posts, two proud-moment posts, and one personal post per week to balance authority with relatability. It has nothing to do with lookalike or predictive audiences.
Does Meta still have lookalike audiences?
Yes. Meta still offers lookalike audiences built from a source of at least 100 people from one country, and it recommends sources in the low thousands. LinkedIn is the platform that discontinued them, so cross-channel advertisers now run two different similarity systems.
It’s Time to Stop Renting Your Audience
LinkedIn ended lookalike audiences, and honestly? It forced a healthier question. Not “which feature replaces the old one,” but “which audiences do I actually own.”
So run the ads-side alternatives where they earn their keep. And build the list that’s yours. Picture it: one clean seed of your best customers, 25 fresh names per champion, sitting in your CRM where no deprecation notice can touch them. That’s a durable answer to a platform-shaped problem.
You got this. Tell me in the comments: did the February 2024 change catch your campaigns off guard, and what did you switch to?
Bad
Okay
Good
Amazing
