DV360 Lookalike Audiences: A Practical Guide to Scaling Prospecting

What Is a Lookalike Audience in DV360?

A lookalike audience — often labelled a similar audience within Google's ecosystem — is a modelled segment of new users who share behavioural and contextual traits with an existing group you already value. In Display & Video 360, you provide a seed audience (your best customers, high-intent visitors, or a first-party list), and the platform's modelling identifies users who resemble that seed but haven't yet engaged with your brand.

The goal is simple: take the signal locked inside your existing data and use it to find more people like the ones already converting. Done well, lookalikes bridge the gap between narrow retargeting and broad, wasteful prospecting.

Lookalike vs. remarketing vs. broad prospecting

It helps to place lookalikes on a spectrum of intent and scale:

Audience type	Primary use	Typical scale	Intent signal
Remarketing	Re-engage known users	Small	High
Lookalike / similar	Find new users like your best ones	Medium	Modelled
Broad prospecting	Top-of-funnel reach	Large	Low

Lookalikes sit in the middle — they extend reach beyond your owned audiences while keeping targeting anchored to a quality signal.

Why Lookalikes Matter for DV360 Buyers

Most advertisers exhaust their remarketing pools quickly. Once you've spoken to everyone who visited your site or opened your app, growth stalls unless you can reliably find new prospects. Lookalike modelling is the mechanism that turns a finite first-party asset into a scalable acquisition engine.

Key benefits include:

• Efficient scale. You reach fresh users without abandoning the quality signal that makes your seed valuable.
• Better than demographic guessing. Modelled similarity outperforms broad age/interest assumptions in most accounts we manage.
• First-party data activation. Your CRM lists, purchase data, and high-value site actions become the foundation for growth rather than sitting idle.

Because DV360 connects tightly to the wider Google Marketing Platform, audiences built from your Google Analytics 4 property, Ads Data Hub segments, or Customer Match lists can flow into your line items with minimal friction.

Building a Seed Audience That Actually Performs

The quality of a lookalike is almost entirely determined by the quality of its seed. Feed the model noisy, low-intent data and you'll get a large but hollow audience. The discipline is in the seed.

Choose the right source

Strong seed candidates include:

• Purchasers or converters — not just any visitor, but users who completed a valuable action.
• High-value customer segments — repeat buyers, high-LTV cohorts, or subscription retainers from your CRM.
• Deep engagement events — users who reached a pricing page, added to cart, or completed onboarding.
• YouTube and video engagers — where you're running CTV or in-stream, engaged viewers make useful seeds.

Avoid seeding from bounced traffic, all-site visitors, or lists padded with unqualified leads. A smaller, cleaner seed almost always beats a larger, diluted one.

Get the seed size right

Modelling needs enough examples to find meaningful patterns, but bigger isn't automatically better. Very large seeds can blur the distinct signal you're trying to replicate. As a rule of thumb, aim for a seed that is genuinely homogeneous — a group that shares a real, describable behaviour — rather than simply the biggest list you can assemble.

Keep it fresh

User behaviour drifts. Refresh CRM-based seeds on a regular cadence and rely on membership durations that keep site-based seeds current. Stale seeds model yesterday's customer, not tomorrow's.

Setting Up Lookalikes in DV360

While exact menu labels evolve, the workflow is consistent:

1. Create or import your seed. Build a first-party audience in DV360, activate a GA4 audience, or import a Customer Match list. Confirm the seed meets minimum size and consent requirements.
2. Generate the similar/lookalike segment. DV360 models comparable users based on the seed's characteristics.
3. Assign the audience to a line item. Apply it to a dedicated prospecting line item so you can measure it cleanly.
4. Layer sensible controls. Add brand-safety, geographic, and frequency settings — modelling finds the who, but you still control the where and how often.

For teams standing up this workflow for the first time, our DV360 managed services team routinely handles seed hygiene, audience architecture, and measurement scaffolding so campaigns launch clean.

Structuring Campaigns for Clean Measurement

Lookalikes only prove their worth if you can isolate them. A common mistake is stacking similar audiences on the same line items as remarketing or broad targeting, which makes attribution impossible.

Recommended structure:

• Separate line items for remarketing, lookalike prospecting, and broad reach.
• Exclusion lists to prevent your existing customers and site visitors from appearing in prospecting delivery.
• Distinct naming conventions so reporting rolls up by strategy, not just by creative.

This separation lets you compare cost-per-acquisition and post-view behaviour across strategies and shift budget toward what works.

Layering for precision or scale

Once the base is running, you can tune:

• Tighten by combining a lookalike with contextual or affinity signals when you need higher intent.
• Widen by increasing the modelled reach or adding multiple seeds when volume is the priority.

Start tighter, prove efficiency, then expand. It's far easier to scale a working audience than to salvage a bloated one.

Common Pitfalls to Avoid

• Seeding from low-quality data. The single biggest cause of underperformance.
• Forgetting exclusions. Without them, you pay prospecting rates to reach people you already own.
• Judging too early. Modelled audiences need a fair learning window and sufficient volume before you draw conclusions.
• Ignoring consent and data governance. First-party lists must be collected and used compliantly; build this into your process from day one.
• Set-and-forget seeds. Refresh cadence matters as much as the initial build.

Measuring Lookalike Performance

Judge lookalike prospecting on acquisition metrics, not remarketing benchmarks. Because these users are new to your brand, expect longer paths to conversion and lean on:

• Cost per new customer or qualified lead
• Incremental reach versus your existing audiences
• Post-view and post-click conversion trends over a full consideration window
• Downstream value (LTV, repeat purchase) where your data allows

Comparing a lookalike line item's efficiency directly against a bottom-funnel remarketing line will always flatter the latter — measure each against its job.

Where Lookalikes Fit in Your Wider Strategy

Lookalike audiences are a connective layer, not a standalone tactic. They work best when your remarketing is capturing intent, your broad prospecting is building awareness, and your first-party data is well-organised and consented. If your data foundation is thin, invest there first — modelling can only amplify the signal you give it.

Teams that want hands-on control while retaining expert guidance often opt for a co-managed services arrangement, while those seeking full delegation lean on our broader DV360 services. Either way, the audience strategy should map to genuine business outcomes, not vanity reach.

Ready to Scale Smarter Prospecting?

Lookalike audiences reward advertisers who treat their first-party data as a strategic asset. If you'd like help auditing your seeds, structuring prospecting line items, or building a measurement framework that proves incrementality, talk to our DV360 experts — we'll help you turn your best customers into your next ones.