Finding New Customers with Lookalike Audiences: When It Works, When It Doesn't

For years, the standard Meta ads recipe was simple: upload your customer list, build a 1% lookalike, launch the campaign. It was straightforward — and it worked. Then Meta overhauled its ad delivery system in late 2025, and that recipe started showing its age. But don't throw it out just yet. Under the right conditions, lookalike audiences are still one of the strongest cards on the table.

How Lookalike Audiences Work in 2026

As of October 2025, Meta rebuilt how its ad system works. In the old model, you picked an audience and Meta showed them your ad. In the new model, the creative itself — the image, copy, tone of your ad — also shapes who sees it. According to Meta's own data, more than half of campaign performance now comes down to creative quality. That means even a perfectly built lookalike audience can underperform if the ad itself is weak. On top of that, what was previously called 'Lookalike Expansion' has been rebranded as 'Advantage+ Lookalike' — and it no longer acts as a hard boundary. Think of it as a hint you give to the system; Meta looks at your list, learns from it, and may expand into better-converting segments it finds on its own.

When Does It Actually Work?

• If your account generates fewer than 50 conversions per week: Meta's automated targeting system needs time to gather enough data. Below this threshold, a well-built 1% lookalike can deliver meaningfully lower acquisition costs compared to broad targeting.
• If your monthly budget isn't enough to complete the learning phase: The Advantage+ system needs at least 50 results per week. If your budget doesn't support that, a lookalike audience offers a more predictable alternative.
• If you're expanding into a new city or region: A lookalike built from your existing customer data gives you a reasonable starting point in unfamiliar geography.
• If you're in B2B or a niche industry: In sectors like machinery, manufacturing, or professional services — where you have a small but high-quality customer list — lookalike targeting can significantly outperform interest-based targeting on cost.

When Are You Fooling Yourself?

• If your account produces 50+ conversions per week and you have Conversions API set up: Under these conditions, Meta's automated targeting consistently outperforms lookalike audiences — and the cost difference is not negligible.
• If your seed list is low-quality or mixed: A list of 500 real buyers produces a far stronger lookalike than 10,000 generic email subscribers. Including everyone dilutes your results.
• If you're a new brand with no conversion data: Without a seed list, you can't build a lookalike. At this stage, collect data first with broad targeting.
• If you're starting with a wide percentage like 5-10%: Test with 1% first, then expand based on performance. Starting broad dilutes audience quality from the start.
• If you're running it as 'set and forget': Never leave a lookalike running on its own. Run regular A/B tests alongside Advantage+ to see what's actually performing.

A Practical Approach for Turkish SMBs

Most small and medium businesses in Turkey outside of large e-commerce players fall below the 50 weekly conversions threshold. This keeps lookalike audiences relevant — but only when a few prerequisites are in place. Without Conversions API integration, the data you send to Meta is incomplete, and an incomplete seed produces an incomplete lookalike. Without clean, segmented customer data, your seed list is mostly noise. For local service businesses, a 1% lookalike combined with city-level geographic targeting remains a practical, actionable strategy.

The bottom line: lookalike audiences aren't dead — they've just changed. Used under the right conditions, with the right seed and a strong creative, they still deliver. But a lookalike set up carelessly, left untouched, and run in isolation will quietly drain your budget. Knowing the difference is what makes the difference.