Lookalike Audiences: How to Find More Customers Who Resemble Your Best Ones

Lookalike Audiences: How to Find More Customers Who Resemble Your Best Ones

What Are Lookalike Audiences and Why They Matter

Lookalike audiences allow you to reach new people who share characteristics with your existing customers. Instead of guessing which demographics or interests might correlate with purchase intent, you provide the ad platform with a list of your best customers and let its algorithms find more people like them. This lookalike audience guide covers how to build, deploy and optimise these audiences across Google Ads and Meta Ads for maximum impact.

The value proposition is straightforward: your existing customers share behavioural and demographic patterns that predict purchase intent. By identifying those patterns at scale, ad platforms can find prospects who are statistically more likely to convert than a random audience. This makes lookalike targeting one of the most efficient prospecting tools available.

For Singapore businesses, lookalike audiences are particularly valuable because the domestic market is small enough that broad targeting wastes significant budget on irrelevant impressions. A lookalike audience guide approach lets you focus spend on people who genuinely resemble your buyers, whether you are targeting Singapore specifically or expanding into regional ASEAN markets where you have less first-hand market knowledge.

How Lookalike Audiences Work Across Platforms

Both Meta and Google use machine learning to analyse the characteristics of your source audience and find new users with similar profiles. The specific signals differ by platform. Meta analyses demographic data, interests, behaviours and social connections. Google analyses search behaviour, browsing history and app usage.

The process starts with a source audience, sometimes called a seed audience. This is a list of your existing customers, website visitors, app users or engaged contacts. The platform analyses this group, identifies common traits and then finds new users who share those traits but are not already in your source list.

Most platforms let you control the size of your lookalike audience. A smaller percentage, like 1 per cent on Meta, produces a highly similar audience with a narrow reach. A larger percentage, like 5-10 per cent, broadens the reach but reduces similarity to your source. The right size depends on your campaign objectives: prospecting campaigns may benefit from broader lookalikes, while high-value campaigns perform better with narrow ones.

It is worth noting that Google retired its standalone Similar Audiences feature in 2023 and moved towards audience signals within Performance Max and other campaign types. The underlying technology still leverages your customer data to find similar users, but the implementation has shifted. We will cover both Meta’s explicit lookalike system and Google’s current approach in the sections below.

Building Your Source Audience

The quality of your lookalike audience depends entirely on the quality of your source audience. Garbage in, garbage out. Your source list should represent the customers you most want to replicate, not just anyone who has ever interacted with your business.

Start with your highest-value customers. Upload a list of customers who have made repeat purchases, spent above your average order value or remained loyal over time. This tells the algorithm to find people who resemble your best customers, not your average ones. A customer list of your top 20 per cent by lifetime value typically produces the strongest lookalikes.

Ensure your source audience is large enough for the algorithm to identify patterns. Meta recommends a minimum of 1,000 contacts, though 5,000 to 10,000 produces more reliable results. For Google, larger source audiences give the system more data to work with when generating audience signals.

Segment your source audiences by value proposition. If you sell multiple products or services, create separate source audiences for each. A lookalike based on customers who bought Product A will differ from one based on Product B customers. This segmentation lets you create tailored campaigns for each product line. Upload clean, formatted data that matches the platform requirements. Include email addresses, phone numbers and any additional identifiers the platform accepts to improve match rates. Work with your Google Ads team to maintain and refresh these lists regularly.

Lookalike Audiences on Meta (Facebook and Instagram)

Meta’s Lookalike Audiences remain one of the platform’s most powerful targeting tools. Create them in Meta Ads Manager by selecting a source audience and choosing a target country and audience size percentage. For Singapore campaigns, a 1 per cent lookalike typically includes around 30,000 to 50,000 people, which is sufficient for most campaigns.

Use Custom Audiences as your source. Options include customer lists uploaded from your CRM, website visitors tracked via the Meta Pixel, app users, video viewers and people who have engaged with your Facebook or Instagram content. Each source type produces different lookalike characteristics, so test multiple sources to find the best performers.

Layer additional targeting on top of your lookalikes with caution. Adding interest or demographic filters to a lookalike audience narrows it further, which can improve relevance but also limits the algorithm’s ability to find optimal delivery. For initial testing, run lookalikes without additional restrictions. Add layers only if performance data suggests the broad lookalike includes too many irrelevant users.

Create multiple lookalike audiences at different percentages and test them against each other. Run separate ad sets for 1 per cent, 3 per cent and 5 per cent lookalikes with the same creative and budget. The results will show you the optimal balance between reach and similarity for your specific business. Feed these insights into your broader Facebook advertising strategy.

Similar Audiences and Signals on Google Ads

Google’s approach to audience expansion has evolved significantly. The standalone Similar Audiences feature was sunset in 2023. In its place, Google now uses audience signals within Performance Max campaigns and optimised targeting across other campaign types to achieve similar outcomes.

In Performance Max, you provide audience signals that tell Google who your ideal customers are. These signals include Customer Match lists, website visitor audiences and custom segments based on search terms and URLs. Google’s algorithm uses these signals as starting points to find new users who are likely to convert, effectively creating a dynamic lookalike that updates in real time.

For Search and Display campaigns, optimised targeting expands your reach beyond your defined audiences to find users who are likely to convert based on your campaign’s performance data. This feature uses your conversion data to identify patterns and find new audiences automatically. You can enable or disable optimised targeting at the ad group level.

Customer Match remains the most direct way to leverage your existing customer data on Google. Upload your customer list, and Google matches it against its user base. While Google no longer creates a separate “similar” audience from this list, the data feeds into its broader targeting algorithms. Ensure your Customer Match lists are regularly updated and contain at least 1,000 matched users for effective use as audience signals.

Optimising Lookalike Audience Performance

Refresh your source audiences regularly. Customer behaviour changes over time, and a source audience based on data from two years ago may not reflect your current ideal customer. Update your customer lists quarterly and rebuild lookalike audiences to ensure they reflect your most recent buyer profiles.

Test different source audience definitions. Compare lookalikes based on all customers versus high-value customers. Test lookalikes from purchase data against those from engagement data. Try website visitor lookalikes from different page categories. Each variation targets a slightly different prospect profile, and performance will vary.

Use exclusion audiences to prevent waste. Exclude your existing customer list from lookalike campaigns so you do not pay to reach people who have already converted. Exclude recent website visitors if you have separate retargeting campaigns. These exclusions ensure your lookalike budget goes entirely toward new prospect acquisition.

Monitor frequency and ad fatigue within lookalike audiences. Because these audiences are relatively narrow, especially at lower percentages, creative fatigue sets in faster than with broader targeting. Rotate creative regularly and watch for rising costs per click and declining click-through rates as signals that your audience needs refreshing. Track performance through your digital marketing dashboard alongside other acquisition channels.

Common Mistakes to Avoid

Using too small a source audience is a frequent error. If your source list has only 200 contacts, the algorithm has insufficient data to identify meaningful patterns. The resulting lookalike audience guide recommendations will be unreliable. Wait until you have at least 1,000 matched contacts before creating lookalike audiences.

Including low-quality customers in your source audience dilutes its effectiveness. If your source includes one-time buyers who returned their purchases, customers acquired through deep discounts or contacts who never actually converted, the lookalike will find more people like them. Curate your source list to include only customers you genuinely want more of.

Running the same lookalike audience indefinitely without refreshing leads to audience exhaustion. As the platform repeatedly targets the same pool of users, costs rise and performance declines. Rebuild lookalikes with fresh data every quarter and test new source audiences regularly.

Neglecting creative optimisation while focusing solely on audience targeting is another common mistake. Even the most precisely targeted audience will not convert if your ad creative is weak. Pair your lookalike strategy with strong, tested creative that speaks to the needs and motivations of your target audience. Creative and targeting work together, and neither can compensate for deficiencies in the other.

Frequently Asked Questions

What is the minimum source audience size for lookalike audiences?

Meta requires at least 100 source contacts but recommends 1,000 to 5,000 for reliable results. Google’s Customer Match requires 1,000 matched users. In practice, source audiences of 5,000 or more produce the most consistent lookalike performance.

What percentage lookalike should I start with?

Start with 1 per cent for the highest similarity to your source audience. This is typically the best-performing option for conversion-focused campaigns. Test wider percentages of 3-5 per cent for awareness campaigns where reach is more important than precision.

Can I use lookalike audiences for a Singapore-only campaign?

Yes. On Meta, select Singapore as the target country when creating your lookalike. The 1 per cent audience will be drawn from Singapore’s user base. For Google, set your campaign location targeting to Singapore and provide audience signals that guide the algorithm toward local users.

How often should I refresh my lookalike audiences?

Refresh source audiences and rebuild lookalikes quarterly at minimum. If your business is growing rapidly or your customer profile is shifting, monthly refreshes may be warranted. Dynamic sources like website visitor audiences update automatically.

Do lookalike audiences work for B2B campaigns?

Yes, though the source audience must reflect your B2B buyer profile. Upload lists of decision-makers rather than company names. On LinkedIn, similar audience features are specifically designed for B2B targeting. On Meta, B2B lookalikes work best when combined with job title or industry filters.

What data should I include in my customer list upload?

Include email addresses, phone numbers, first and last names, city and country. More data points improve match rates. Hash personal data before uploading to Meta and Google for privacy compliance. Both platforms handle hashing automatically if you use their upload tools.

Can I use website visitors as a source for lookalike audiences?

Yes. Both Meta and Google allow you to create lookalikes from website visitor audiences. For best results, use visitors who completed high-intent actions like adding to cart, starting checkout or viewing pricing pages rather than all website visitors.

How do lookalike audiences compare to interest-based targeting?

Lookalike audiences typically outperform interest targeting for conversion campaigns because they are based on actual customer data rather than assumed correlations. Interest targeting can be useful for awareness campaigns or when you lack sufficient customer data for lookalikes.

Why are my lookalike audiences not performing well?

Common causes include poor source audience quality, exhausted audiences that need refreshing, weak ad creative, landing page issues and insufficient budget for the audience size. Diagnose by testing different source audiences, refreshing creative and checking your conversion funnel.

Are lookalike audiences affected by privacy changes?

Yes. iOS privacy changes and cookie deprecation have reduced the data available for building and matching audiences. Use first-party data like customer email lists as source audiences rather than relying solely on pixel-based website visitor audiences. Server-side tracking and Conversions API implementation help maintain data quality.