If you run Demand Gen campaigns, you have less than a month to fix your audience lists. Starting April 30, 2026, Google enforces a uniqueness check on all Lookalike user lists — meaning overlapping or duplicate customer data across your segments will either get flagged or underperform. Most PPC managers don't realize their lists already have this problem.

This guide walks you through the new Lookalike Segments system that replaced Similar Audiences, how to structure first-party data lists without duplicates, and the combination strategy with NCA Mode that's delivering an 11.5% improvement in new-to-returning customer ratios. If you're running Demand Gen without addressing this, you're paying for redundant reach.

What Changed: From Similar Audiences to Lookalike Segments

Google officially discontinued Similar Audiences in August 2023 and replaced them with Lookalike Segments inside Demand Gen campaigns. The shift wasn't cosmetic. The underlying model changed fundamentally.

Similar Audiences worked by mirroring your remarketing lists — Google found users who looked like your website visitors or converters and built static audiences from that data. The problem was transparency. You couldn't control how broadly Google expanded the match, and list quality degraded over time as the seed audience evolved but the similar audience didn't.

Lookalike Segments fix this with three explicit controls: narrow, balanced, and broad targeting options. Narrow focuses on the top 2.5% of users most similar to your seed list. Balanced extends to roughly 5%. Broad pushes to 10% and beyond. You choose the trade-off between precision and reach directly.

But the bigger change is data sourcing. Lookalike Segments in Demand Gen pull from Customer Match lists, app user lists, and website visitor lists — not just remarketing audiences. That means your first-party data quality directly determines Lookalike performance. Bad input data means bad expansion. Clean, deduplicated lists mean more qualified reach.

The April 30, 2026 uniqueness check makes this even more critical. Google will now validate that the user lists you feed into Lookalike Segments contain unique entries. Duplicate users across multiple seed lists — or within a single list — will trigger warnings and potentially limit segment creation.

Takeaway: Lookalike Segments give you more control than Similar Audiences ever did, but they demand cleaner data. The narrow/balanced/broad options are only as good as the seed list behind them.

The April 30 Uniqueness Check: What It Means for Your Campaigns

Here's the specific change every PPC manager needs to understand. Starting April 30, 2026, Google will enforce a uniqueness validation on user lists used for Lookalike Segments in Demand Gen campaigns. This means the system checks whether individual users appear in multiple lists submitted to the same campaign or ad group.

Why does this matter? Because most advertisers built their audience strategy by creating overlapping lists. You might have a "High-Value Purchasers" list, an "All Converters" list, and a "Repeat Buyers" list — and many users exist in all three. Before the uniqueness check, Google ignored this overlap and simply expanded from each list independently. After April 30, those overlapping entries reduce the effective seed size and can trigger list rejection.

The practical impact hits three areas.

List creation will fail. If you try to create a Lookalike Segment from a list with excessive internal duplicates, the system may reject it entirely or flag a warning that reduces its serving priority.

Campaign performance will drop. Overlapping seed lists mean Google's expansion model sees the same user signals multiple times, skewing the lookalike profile. Instead of finding genuinely new prospects, the algorithm over-indexes on characteristics of your most duplicated users.

Budget efficiency suffers. When multiple Lookalike Segments in the same campaign share seed users, you're bidding against yourself for the expanded audiences. Frequency climbs. Cost per new customer rises.

The fix isn't complicated, but it requires a structured audit of every user list you're feeding into Demand Gen. And you need to do it before April 30.

Takeaway: The uniqueness check isn't optional. Audit your user lists now for cross-list and within-list duplicates. After April 30, overlapping lists will hurt your Lookalike Segment performance directly.

---

Are your campaigns healthy? Duplicate audience lists are just one of dozens of hidden issues that drain budget in Demand Gen campaigns. AdsHealth uses AI to diagnose your Google Ads campaigns and shows you exactly where you're leaving money on the table. Get your free report →

---

How to Structure Deduplicated User Lists for Lookalike Segments

The goal is simple: each user list should contain a distinct, non-overlapping group of customers that represents a clear behavioral signal. Here's the framework.

Step 1: Export and merge all existing lists. Pull every Customer Match list, conversion-based list, and website visitor list you're currently using in Demand Gen. Merge them into a single master file with a unique identifier per user (hashed email or phone number).

Step 2: Define mutually exclusive segments. Instead of overlapping lists like "All Buyers" and "High-Value Buyers," create segments where each user appears exactly once based on their highest-value behavior:

• Tier 1 — Top Spenders: Users in the top 10% by lifetime value. These become your narrow Lookalike seed.
• Tier 2 — Regular Converters: Users who've purchased 2+ times but fall outside Tier 1. Balanced Lookalike seed.
• Tier 3 — One-Time Buyers: Single-purchase users. Broad Lookalike seed or exclusion list.
• Tier 4 — Engaged Non-Buyers: Newsletter subscribers, app users, high-intent site visitors who haven't purchased.

Step 3: Validate uniqueness before upload. Run a deduplication check across all four tiers. No user should appear in more than one list. Google's system will catch overlaps after April 30 — it's better to clean them proactively.

Step 4: Match Lookalike expansion to tier quality. Your best data (Tier 1) can handle narrow targeting because the behavioral signals are strong and consistent. Lower tiers benefit from broader expansion because the individual signals are weaker — Google needs more room to find patterns.

This tiered approach also solves a common Demand Gen audience targeting strategy problem: knowing which Lookalike width to use. The answer depends on seed quality. High-quality, homogeneous seed lists perform best at narrow. Diverse or noisy lists need balanced or broad to find workable patterns.

If you're already managing first-party data strategy across your ad accounts, this framework extends naturally. The same data hygiene principles apply — just with the added constraint of cross-list uniqueness.

Takeaway: Build mutually exclusive user tiers based on value. Each user appears in exactly one list. Match Lookalike width (narrow/balanced/broad) to the quality and homogeneity of that tier's data.

Combining Lookalike Segments with NCA Mode for Maximum New Customer Reach

Here's where the Google Ads similar audiences replacement gets genuinely powerful. Demand Gen campaigns now support New Customer Acquisition (NCA) Mode, and when combined with Lookalike Segments, the results are measurable.

NCA Mode tells Google's bidding algorithm to prioritize users who have never interacted with your brand. You provide your existing customer list as an exclusion signal, and Google adjusts bids to favor new prospects. This works with any Demand Gen audience, but it's especially effective with Lookalike Segments.

Why? Because Lookalike Segments already target users who resemble your best customers. Adding NCA Mode on top ensures you're not wasting that expansion on people who already bought from you. The combination eliminates the most expensive form of waste in prospecting campaigns: paying to "acquire" existing customers.

The data backs this up. Advertisers running Lookalike Segments with NCA Mode report an 11.5% improvement in new-to-returning customer ratio compared to Lookalike alone. That's not a marginal gain — for a brand spending $50,000/month on Demand Gen, an 11.5% shift toward genuinely new customers represents thousands of dollars in incremental value.

The setup is straightforward:

1. Upload your complete customer list as a Customer Match audience (this becomes the NCA exclusion signal).
2. Create separate, deduplicated seed lists for your Lookalike Segments (using the tiered approach from the previous section).
3. Enable NCA Mode at the campaign level.
4. Set your NCA bid adjustment based on how much more a new customer is worth versus a returning one.

One critical detail: the customer list used for NCA exclusion should be comprehensive — every known customer, not just recent ones. The Lookalike seed lists should be selective and segmented. These are two different data jobs.

Are you running Lookalike Segments without NCA Mode? You're likely paying to re-acquire 15-25% of users who already know your brand.

Takeaway: NCA Mode + Lookalike Segments is the highest-leverage combination in Demand Gen. Enable NCA, upload a comprehensive exclusion list, and watch your new customer acquisition cost drop.

---

Stop guessing what's wrong with your ads. If your Demand Gen campaigns are spending budget on existing customers instead of new prospects, you need visibility into what's actually happening. AdsHealth gives you an AI-powered health score and actionable recommendations in minutes. Free diagnosis →

---

Customer Match + Lookalike: The Most Effective Seed Source

Not all seed lists produce equal Lookalike Segments. The data is clear: Customer Match lists consistently outperform other seed sources for Lookalike expansion quality in Demand Gen campaigns.

The reason is signal density. Customer Match lists contain hashed first-party identifiers — emails, phone numbers, addresses — that Google can match directly to logged-in user profiles. Website visitor lists rely on cookies and consent signals that have degraded significantly since third-party cookie restrictions tightened. App user lists are strong but limited to brands with significant app adoption.

Customer Match gives Google the cleanest identity signal possible. When the Lookalike model expands from a Customer Match seed, it's working with verified user profiles rather than probabilistic browser signals. The resulting Lookalike Segment contains users whose verified behaviors and interests genuinely resemble your customers — not users whose anonymous browsing patterns happen to look similar.

Here's the practical hierarchy for seed source effectiveness:

1. Customer Match (hashed emails from CRM) — Highest match rate, strongest behavioral signals, best Lookalike quality.
2. Customer Match (phone numbers) — Slightly lower match rate but still strong identity signals.
3. App user lists — Good signals but limited scale for most advertisers.
4. Website visitor lists (GA4 audiences) — Weakest identity signal, most affected by consent and cookie changes.

For the first party data lookalike Google strategy to work at maximum effectiveness, invest in growing and maintaining your Customer Match lists. That means collecting first-party data at every touchpoint: checkout, newsletter signup, account creation, loyalty programs. Every verified email that enters your CRM improves your Lookalike seed quality.

If you've already built a retargeting and remarketing strategy, your Customer Match lists likely exist but may need cleaning. Remove bounced emails, inactive accounts, and users who've opted out. A smaller, cleaner Customer Match list produces better Lookalike Segments than a large, noisy one.

Takeaway: Prioritize Customer Match lists as your primary Lookalike seed source. Invest in first-party data collection. Clean your CRM data before uploading — list quality beats list size every time.

Narrow vs. Balanced vs. Broad: Choosing the Right Lookalike Width

The three Lookalike expansion options in Demand Gen aren't just a slider between "small audience" and "big audience." Each serves a distinct strategic purpose, and choosing wrong wastes budget.

Narrow (top 2.5% similarity) works best when your seed list is large (10,000+ users), homogeneous (similar purchase behavior or value), and you're optimizing for cost-per-acquisition. Narrow Lookalikes produce smaller audiences with higher conversion rates. Use this for bottom-funnel Demand Gen campaigns where you need qualified leads, not volume.

Balanced (approximately 5% similarity) is the default and the right choice for most campaigns. It gives Google enough room to find patterns without diluting the audience beyond usefulness. If you're unsure, start here. Balanced works well with mid-size seed lists (5,000-15,000 users) and standard CPA or ROAS bidding strategies.

Broad (10%+ similarity) is for awareness and top-funnel reach. The audience quality drops compared to narrow and balanced, but the scale increases significantly. Use broad when you're optimizing for reach, brand awareness, or when your seed list is too small (under 5,000 users) for narrow or balanced to generate sufficient volume.

The mistake most managers make is running all three widths from the same seed list in the same campaign. This creates internal competition — your narrow, balanced, and broad Lookalikes are all expanding from identical data, and the broader segments fully contain the narrower ones. Google's auction system handles some of this, but you're still fragmenting your own budget.

Instead, match width to seed tier as described in the structuring section above. Your highest-value customer tier gets narrow. Your mid-tier gets balanced. Your broadest tier gets broad. Each width expands from a different data profile, producing genuinely distinct audiences.

How aligned is your Demand Gen audience targeting strategy with the actual quality of your seed data? If you're running broad Lookalikes from your best customers, you're under-leveraging your strongest data asset.

Takeaway: Match Lookalike width to seed list quality and campaign objective. Don't run multiple widths from the same list — use distinct, tiered seed data for each expansion level.

Your Pre-April 30 Action Plan

The uniqueness check deadline is real and the clock is ticking. Here's what to do in the next three weeks.

Week 1: Audit and Export. Pull every user list currently active in your Demand Gen campaigns. Export them. Cross-reference for duplicates — both within individual lists and across lists. Document the overlap percentage. Most advertisers find 20-40% duplication across their audience lists.

Week 2: Restructure and Deduplicate. Build the tiered, mutually exclusive list structure described in this guide. Remove all cross-list duplicates. Upload clean Customer Match lists for each tier. Set up your comprehensive NCA exclusion list separately.

Week 3: Reconfigure and Monitor. Update your Demand Gen campaigns with the new Lookalike Segments. Match narrow/balanced/broad widths to the appropriate tiers. Enable NCA Mode. Run for 7-14 days before the deadline to establish baseline performance before the uniqueness check enforcement changes the landscape.

The advertisers who treat this as a "set it and forget it" task will struggle. Audience list hygiene is ongoing — as customers move between tiers (one-time buyers become repeat buyers, for example), your lists need updating. Build a monthly refresh cadence into your campaign management workflow.

If your Demand Gen campaigns are already underperforming, the audience list structure is likely a contributing factor. The uniqueness check will either fix the problem (by forcing cleaner data) or expose it (by breaking campaigns built on overlapping lists). Either way, proactive restructuring puts you ahead.

Takeaway: Start the audit now. Three weeks is enough time to restructure your lists if you begin immediately. After April 30, cleaning up retroactively will cost you campaign downtime and wasted budget.

---

Find out what's killing your ROAS. Your Demand Gen audiences might be overlapping, your NCA Mode might be misconfigured, or your Lookalike seeds might be pulling from the wrong data. AdsHealth diagnoses your Google and Meta campaigns with AI — and tells you exactly what to fix. Get your free report →