Lookalike Audience Strategy: 37 Ad Tactics for Rapid SaaS Growth

A lookalike audience is an ad targeting option that finds new people who resemble a group you already know, usually your best customers. You give the ad platform a source audience (often called a seed), such as a customer list or your website visitors, and its model finds other users who share their characteristics, so your ads reach likely buyers who aren’t in your source audience.

B2B and SaaS teams use lookalikes to turn their best accounts into a prospecting audience: they build the seed from high-value customers, qualified leads or converted trials, launch the lookalike on Meta or in Google Ads Demand Gen or Video campaigns, and measure it on pipeline rather than clicks. That definition matches how Meta, ZoomInfo and Salesforce describe the feature, and ZoomInfo’s B2B guide adds that you need closed-loop measurement before you judge a lookalike campaign on pipeline. The platforms have changed the feature since 2023, though: Google retired similar audiences and LinkedIn retired lookalike audiences, so check what each platform offers today before you plan a campaign.

Key takeaways

• Meta needs a lookalike source of at least 100 people and generally recommends 1,000 to 5,000, according to its Business Help Center (checked November 2026).
• Google Ads removed similar audiences from all campaigns on August 1, 2023; today Lookalike segments are available only in Demand Gen and Video brand campaigns, at narrow 2.5%, balanced 5% or broad 10% of users in the target location (Google Ads Help, November 2026).
• Google is moving Demand Gen Lookalike segments (not Video brand ones) to a suggestion mode during 2026, where the seed list guides delivery instead of limiting it (Google Ads Help, November 2026).
• LinkedIn discontinued lookalike audiences on February 29, 2024 and points advertisers to predictive audiences, which need a source of at least 300 members (LinkedIn Help, November 2026).
• Seed quality matters as well as size: Meta says you may get better results, depending on your goals, from an audience made from your best customers than from one that includes all your customers (Meta Business Help Center, November 2026).

What is a lookalike audience?

Meta’s own definition is the clearest: a lookalike audience is “a way your ads can reach new people who are likely to be interested in your business because they share similar characteristics to your existing customers,” built from the demographics, interests and behaviors of your source audience (Meta Business Help Center, November 2026). Google calls the same idea Lookalike segments, built from a seed list of your first-party data.

Three terms come up again and again. The seed or source audience is the group you start from. The lookalike is the new audience the platform builds from it. The size or percentage sets how closely the new audience must match the seed: Meta says smaller percentages match the source more closely, while larger ones create a bigger, broader audience.

For B2B companies there is a second kind of lookalike that doesn’t come from an ad platform. Data providers such as ZoomInfo build lookalike lists of companies from firmographic and technographic signals (industry, size, technology used), which sales and account based marketing teams then work directly. ZoomInfo’s guide draws exactly that line: ad platforms model people, B2B data providers model accounts.

Custom audience vs lookalike audience

A custom audience is people you already know: your customer list, your website visitors, people who opened a lead form. You use it to retarget or to exclude. A lookalike audience is people you don’t know yet who resemble a custom audience. Every lookalike starts from a source you already hold, such as a custom audience, and Meta removes the people in the source from the lookalike it builds.

How a lookalike audience works

Every platform runs the same five steps, and the first one decides the rest:

1. Choose the seed. Pick the people you want more of: paying customers, your highest value accounts, qualified leads or trial users who converted.
2. Upload or connect it. Send a customer list, or point the platform at pixel, form or conversion data it already holds.
3. Match. The platform matches your records to its own users. If too few match, the lookalike isn’t created.
4. Model. The platform finds users who share the matched people’s traits, within the size you chose.
5. Target and exclude. You run ads to the new audience and keep existing customers out of it.

Populating takes time. Meta says a new lookalike audience can take up to 3 days to fill, and Google says a new Lookalike segment takes about 72 hours, whether or not a campaign uses it yet.

Which platforms still offer lookalike audiences?

Not all of them. The table shows what each major B2B ad platform offers as of November 2026, according to its own help pages.

Platform	What it offers today	Seed rules	Size and expansion	Source
Meta (Facebook, Instagram)	Lookalike audiences, plus value-based lookalikes and Advantage+ lookalike	At least 100 people in the source; 1,000 to 5,000 recommended; up to 500 lookalikes per source	Percentage slider; Advantage+ lookalike may reach beyond it and is on by default for list, website and app sources	Meta Help, Advantage+ lookalike
Google Ads, Demand Gen and Video brand campaigns	Lookalike segments; the Demand Gen ones are moving to a suggestion mode during 2026	Customer Match list, website, app or YouTube data; up to 10 lists	Narrow 2.5%, balanced 5% or broad 10% of users in the target location; in Demand Gen suggestion mode these act as signals	Demand Gen, Video, suggestion mode
Google Ads, other campaigns	No similar audiences since August 1, 2023; optimized targeting and audience expansion instead	Your first-party segments become signals	Optimized targeting is on by default and looks beyond your segments	Google Ads Help (2023, archived), optimized targeting
LinkedIn	No lookalike audiences since February 29, 2024; predictive audiences and Audience Expansion instead	One source type: list, Lead Gen Form, conversion or retargeting audience; at least 300 members; lists of 300 to 300,000 rows	Predictive audiences refresh daily; Audience Expansion is turned off on ad sets that use them	LinkedIn Help

Meta

Meta still has the classic lookalike. You need a source of at least 100 people, and Meta “generally” recommends 1,000 to 5,000; saved audiences can’t be used as a source, and you can build up to 500 lookalikes from one source (Meta Business Help Center, November 2026). You no longer pick a location when you create one: Meta uses the ad set’s location instead. In Ads Manager you set the size with a slider as a percentage of the people in that location; Meta’s Marketing API documentation accepts 1% to 20%.

Two newer options matter. A value-based lookalike starts from a customer list with a value column, so Meta knows which customers are worth the most (Meta Business Help Center). Advantage+ lookalike, which Meta also describes as using a lookalike audience as a suggestion, lets Meta reach people beyond the percentage you chose when it finds better performance there, and Meta says it is switched on automatically for new campaigns using lookalikes built from mobile, website or customer list sources (Meta Business Help Center, November 2026). One caution for fintech founders: Meta says lookalike audiences are limited or unavailable for some ads about financial products and services, employment or housing.

Google Ads

Google announced on November 1, 2022 that it would retire similar audiences. From May 1, 2023 no new similar segments were generated, and on August 1, 2023 they were removed from all ad groups and campaigns; Display, Discovery and Video action campaigns still using them were moved to optimized targeting (Google Ads Help, archived April 2023). Google’s current advice is to turn on optimized targeting and add your first-party segments as hints.

The exceptions are Demand Gen and Video brand campaigns. There, Lookalike segments are built from a Customer Match list or people who interacted with your website, app or YouTube channel, at three sizes: narrow (2.5% of users in the target location), balanced (5%) or broad (10%); Google’s Video campaigns page lists the same sizes. Through 2026 Google is moving the Demand Gen ones to a suggestion mode, a change it says does not affect Video brand campaigns: the seed list becomes a signal, the size slider stops acting as a hard limit, and the old 100 user minimum no longer applies. Advertisers who want the old behavior can opt out. If you also run search, our Google Ads strategy guide covers the rest of the account.

LinkedIn

LinkedIn discontinued lookalike audiences on February 29, 2024. Existing ones stopped refreshing, and ad sets still using them deliver to a frozen copy (LinkedIn Help, November 2026). LinkedIn now recommends predictive audiences, which combine your source data with LinkedIn’s AI to find members predicted to act like the people in it, and Audience Expansion for attribute targeting. A predictive audience needs one source type with at least 300 members, contact or company lists of 300 to 300,000 rows, and a maximum of 200 per ad account. LinkedIn notes that a predictive audience built from a list can include the list itself, so exclude it if you only want new people. Our LinkedIn ads guide covers campaign setup.

Do lookalike audiences still work?

Yes, but they work differently than they did a few years ago. The big platforms have moved from treating your seed as a fence to treating it as a hint. Meta’s Advantage+ lookalike can reach past the percentage you picked, Google’s Demand Gen suggestion mode uses the seed list as a signal, and Google replaced similar audiences with optimized targeting, which looks for converters beyond your segments.

That makes the seed more important, not less. When the platform decides how far to go, the quality of the people you hand it is the main thing you still control. A seed of paying customers teaches the model to find buyers; a seed of every newsletter subscriber teaches it to find newsletter readers. None of the platforms publishes a guaranteed lift, so treat any lookalike as a test against your current prospecting audience and judge it on cost per qualified lead or customer.

How B2B and SaaS teams build lookalike audiences: 12 tactics

These are the practices that hold up against the platforms’ own documentation, grouped by step.

Step 1: Build the seed

1. Seed from your best customers, not all of them. Meta says an audience made from your best customers may get better results than one made from everyone. For SaaS that usually means customers above your median contract value or with the strongest retention.
2. Pick one segment first. A seed mixing very different buyers gives the platform a blurry target. Start with the segment that buys fastest and most often.
3. Seed on value when you can. Meta’s value-based lookalike takes a customer list with a value column, such as revenue per customer, so the model leans toward your most valuable accounts rather than treating every customer alike.
4. Use the source type the platform reads best. Meta accepts customer lists, pixel, app and page sources; Google Demand Gen takes Customer Match lists and website, app or YouTube activity; LinkedIn predictive audiences take one of contact or company lists, Lead Gen Forms, conversions or retargeting audiences.

Brian Yam, Head of Marketing at Paragon, gave this advice on choosing which segment to target first in a 2024 talk on LinkedIn ads:

Pick your best segment. Look at your sales data. Look at where you have the fastest sales cycles, where you have the highest win rates and highest conversion rates throughout the funnel, and start with your best one, where you have the best product market fit, because it’s going to deteriorate over time as you go into your longtail segments.

Brian Yam, Head of Marketing, Paragon ($3.5M revenue). Watch at 13:35

Step 2: Prepare the data

5. Clear each platform’s minimum. Meta needs 100 people in the source, LinkedIn predictive audiences need 300 members, and Google Video brand Lookalike segments need more than 100 active matched people across your seed lists (Google says that minimum no longer applies to Demand Gen segments in suggestion mode). Plan for fewer matches than rows: only records the platform can match to a user count.
6. Refresh the seed. Google refreshes Lookalike segments every 1 to 2 days from your data and LinkedIn refreshes predictive audiences daily, so a synced customer list keeps the lookalike current. A one-off upload slowly goes stale.
7. Upload only data you may use. Customer lists are personal data. Check that your privacy notice and the platform’s customer list terms cover the upload before you send it.

Step 3: Build and size the audience

8. Start narrow, then widen. Smaller sizes match the seed more closely: 1% on Meta or narrow (2.5%) on Google. Widen only when the narrow audience stops delivering at a cost you accept.
9. Exclude customers and open deals. Meta drops the source from the lookalike, but your other customer lists and current opportunities are not excluded automatically, and LinkedIn recommends excluding the seed list from predictive audiences if you only want new people.
10. Decide how much expansion you want. Advantage+ lookalike on Meta and suggestion mode on Google can reach beyond your size setting. Leave them on if you optimize for conversions. If you need a clean test of one audience, Google lets you opt out of suggestion mode through a form, while Meta switches Advantage+ lookalike on automatically for new campaigns using list, website or app lookalikes and its help page doesn’t describe turning it off.

Step 4: Launch and measure

11. Write for a cold audience. Most people in a lookalike likely don’t know you yet. Lead with the problem your best customers had, not with a demo request, and move people who engage into a retargeting audience.
12. Judge it on pipeline. Compare cost per qualified lead and customer acquisition cost against your other prospecting audiences, using the same attribution window. Our CAC calculator works out the cost per customer.

Lookalike accounts for account based marketing

Ad platform lookalikes model people. Many B2B teams sell to companies, so they build lookalike accounts instead: a list of companies that resemble their best customers on industry, size, technology and buying signals, which they then target with ads, email and sales outreach. That is the core of account based marketing, and it is what intent and data platforms sell.

Asked in a February 2024 Founderpath interview whether customers upload their best customer list so 6sense can generate a lookalike audience, or whether it identifies the companies browsing their website, Jason Zintak, then CEO of 6sense, said “it’s a little bit of both” and described the data behind it:

We would take a company’s first party data, their CRM system, outbound, or their contacts, known users or known clients and prospects, and their marketing automation, and then marry that with behavioral web data… We’ve mapped five and a half million companies worldwide, which is probably 80 to 85% of what the B2B org needs.

Jason Zintak, CEO, 6sense (February 2024 interview). Watch at 5:12

Watch Jason Zintak’s full interview on how 6sense sells, with transcript. The two approaches combine well: a list of lookalike companies can become a LinkedIn company list, which in turn can seed a predictive audience.

Funding growth without giving up equity

Founderpath provides non-dilutive capital for bootstrapped software founders. Connect your billing, banking and accounting data, and Founderpath shows you what you qualify for, with every figure disclosed up front. There’s no pitch deck, no equity and no board seats.

Check what you qualify for. It takes less than 5 minutes.

Frequently asked questions

What is a lookalike audience?

A lookalike audience is an ad targeting option that finds new people who share the characteristics of a source audience you provide, usually your best customers. The ad platform matches your seed to its users and models others like them, so your ads reach likely buyers who aren’t in your source audience.

What is the difference between a custom audience and a lookalike audience?

A custom audience is people you already know, such as your customer list or website visitors, and is used for retargeting or exclusions. A lookalike audience is new people who resemble a custom audience. Every lookalike is built from a source you already hold, such as a custom audience, and Meta leaves the source’s people out of it.

Can you do lookalike audiences on Google?

Only in Demand Gen and Video brand campaigns. Google removed similar audiences from all campaigns on August 1, 2023, and now offers Lookalike segments in those campaigns at narrow (2.5%), balanced (5%) or broad (10%) sizes. During 2026 Google is moving the Demand Gen ones to a suggestion mode, where the seed list guides delivery rather than limiting it. Other campaign types use optimized targeting instead.

Does LinkedIn still have lookalike audiences?

No. LinkedIn discontinued lookalike audiences on February 29, 2024. It recommends predictive audiences, built from one source such as a contact or company list, Lead Gen Form, conversion or retargeting audience with at least 300 members, and Audience Expansion for attribute targeting.

How big should a lookalike seed audience be?

Above each platform’s minimum, and made of your best customers. Meta needs at least 100 people and generally recommends 1,000 to 5,000. LinkedIn predictive audiences need at least 300 members. Meta says an audience made from your best customers may get better results, depending on your goals, than one that includes all your customers.

What lookalike percentage should I use?

Start with the smallest size, which matches your seed most closely: 1% on Meta or narrow on Google Demand Gen. Widen it only when the small audience stops delivering at an acceptable cost per qualified lead, and compare the sizes as separate tests.

Do lookalike audiences still work in 2026?

Yes, but platforms now use the seed as a signal rather than a hard limit. Meta’s Advantage+ lookalike and Google’s Demand Gen suggestion mode can reach people beyond the size you chose, so the quality of the seed is the main thing you control. Test a lookalike against your current prospecting audience and judge it on pipeline.

What is the best seed audience for B2B lookalike targeting?

Your highest-value customers from your best segment: the accounts with the shortest sales cycles, highest win rates and strongest retention. If you have enough volume, a value-based seed or a list of converted trials or qualified leads also works. Leave churned customers out of the seed, and exclude current customers from the lookalike.

Sources and method

Reviewed November 2, 2026. Platform rules come from Meta’s Business Help Center and Marketing API documentation, Google Ads Help (the 2023 similar audiences notice is linked through the Internet Archive because the live URL now redirects) and LinkedIn’s Marketing Solutions Help, as each stated them on that date; ad platforms change these features often. Founder quotes come from talks published on Founderpath, each linked to the moment it is said, lightly cleaned with fillers removed.

The earlier version of this post listed about 40 founder examples, most of them data or marketing tools that never described running lookalike audiences, with revenue, customer and growth figures that did not show how a lookalike campaign performed. It also described LinkedIn lookalike audiences as current. Those examples, figures and outdated platform claims were removed, along with unsourced FAQ claims that lookalikes convert 3 to 5 times better than cold targeting at 30 to 50% lower cost per acquisition.