Facebook ads agency australia: Why Lookalikes Fail

14 June

If you’re still relying on Lookalike Audiences in Meta, you’re lighting money on fire.

I’ve seen it across dozens of ad accounts we’ve audited in the last year. A 1% Purchase Lookalike that used to be a brand’s top performer is now barely breaking even. A 5% Add-to-Cart Lookalike is bringing in traffic that never converts. The old playbook is broken. It’s not coming back.

When I was scaling my own eCommerce stores, Lookalikes were the closest thing to a magic button we had. Today, that button is disconnected. The brands we see scaling profitably on Meta have moved on. They are using Meta’s own machine learning and their own first-party data in a smarter way. It requires a different way of thinking about audiences, moving from “who are they” to “what did they do”.

The shifting sands of audience targeting post-iOS 14

I remember the pre-2021 days of Meta advertising. Lookalike Audiences were the gold standard for prospecting. You could feed Meta a list of your best 500 customers, ask it to find one million more people just like them, and the algorithm would deliver with incredible precision.

It worked because Meta had a firehose of data. The pixel on your site captured everything, and its connection back to user profiles was frictionless. We could build hyper-accurate audiences based on detailed, reliable signals.

Then Apple released iOS 14.

That single update, with its App Tracking Transparency (ATT) framework, cut the data firehose down to a trickle. When users opted out of tracking, Meta lost visibility. The signals from iPhones and iPads became fuzzy, delayed, or disappeared entirely.

The direct consequence for Lookalike Audiences was catastrophic. To build an accurate lookalike, Meta needs a large, clean seed audience and a massive pool of data to find matches in. iOS 14 damaged both. The seed audiences became polluted with incomplete data, and the prospecting pool became harder to analyse.

We saw the impact almost immediately in the accounts we manage. Lookalike audience sizes shrank. Overlap between different lookalikes increased. Most importantly, the cost per acquisition started to climb. The audiences were simply less precise. Meta was doing its best with less information, but it was like asking a master painter to work in a dark room. You can’t create a masterpiece without light.

This is why a robust Meta CAPI Setup became non-negotiable. It’s a way to send data directly from your server to Meta’s, bypassing the browser and some of the ATT restrictions. It helps, but it doesn’t turn back the clock. The landscape has fundamentally changed. Sticking to the old LAL-first strategy is a recipe for diminishing returns.

Why Meta’s broad targeting often outperforms lookalikes for a facebook ads agency australia

It feels wrong at first. For years, we were taught that granular targeting was the key to success. Now, I’m telling you that for many accounts, the best-performing “audience” is no audience at all.

This is the power of broad targeting.

Meta’s engineers knew the writing was on the wall with data privacy. They have invested billions of dollars into their advertising AI. The algorithm today is vastly more sophisticated than it was five years ago. It no longer needs us to spoon-feed it a perfect audience. It can find your next customer with minimal instruction, provided it has two things: good creative and a clear conversion goal.

When you run a campaign with broad targeting, you are essentially giving the keys to Meta’s machine learning. You define the basics, like country, age, and gender, and then you let the algorithm do the work. It analyses the users who convert and rapidly learns what they have in common, finding pockets of high-intent buyers far more effectively than a stale Lookalike Audience ever could.

This is the engine behind Advantage+ Audience and Advantage+ Shopping Campaigns. You give the system creative assets and a budget, and it finds the most efficient path to conversion. By removing restrictive audience layers, you give the algorithm a bigger playground. It can test different user profiles and follow conversion signals in real-time without being constrained by your assumptions.

We ran a test for a fashion client in Q4 last year. Their 1% Purchase LAL, a long-time top performer, was hitting a 1.8 ROAS. We duplicated the campaign, removed the LAL, and ran it on broad targeting. Within two weeks, the broad campaign was at a 2.9 ROAS. The algorithm, freed from the constraints of the degraded lookalike, found a more profitable audience on its own.

Lookalikes can still have a place in a highly specialised niche or for a brand with very little pixel data. But for most established eCommerce stores, broad targeting is the new starting point for top-of-funnel prospecting.

Our alternative approach: intent-based signals and custom audience layering

While broad targeting is powerful for finding new customers, the real precision in our Meta Ads management comes from leveraging first-party data. We shift the focus from who a person is (demographics) to what they have done (intent).

Your website and customer list are goldmines of these intent-based signals. This is data you own and control. It’s far more reliable than anything Meta can infer about a user’s interests.

An intent-based signal is a specific, high-value action a user takes. These are the building blocks of our most profitable audiences. Examples include: * Added a product to their cart in the last 14 days. * Viewed a specific high-margin product page twice in the last week. * Watched 75% of a key product video. * Initiated checkout but did not purchase. * A customer who has purchased 3+ times. If you’re not sure how to these signals effectively, a Meta Audit can pinpoint exactly where your first-party data strategy can be improved.

The next step is audience layering. This is where we combine these granular custom audiences to achieve extreme precision. For example, we might create a retargeting audience for a skincare brand that includes people who added to cart in the last 30 days, but we’ll exclude anyone who has already purchased in the last 60 days. This prevents us from annoying recent customers with ads for products they just bought.

Another layer could be to target people who visited a specific collection page, like “anti-ageing serums”, and also engaged with the brand’s Instagram profile. This combines on-site behaviour with social engagement to find a highly qualified group of potential buyers.

This approach puts you back in control. You are no longer relying on Meta’s black box. You are building audiences based on the actual behaviour of your users and customers.

Advanced custom audience strategies for a fb ads agency AU

Once you master the basics of intent-based custom audiences, you can build some incredibly effective campaign structures. These are the types of strategies we implement for our clients to scale their accounts.

First, value-based custom audiences. Most eCommerce platforms, like Shopify, allow you to export your customer list with data points like total spend or average order value. We take this data, segment it into tiers (e.g., top 10% of customers by lifetime value), and upload these lists to Meta. This creates a custom audience of your VIPs. You can then use this for targeted campaigns announcing new products first or offering exclusive bundles.

Second, time-based custom audiences are critical. A user’s intent decays quickly. Someone who added a product to their cart yesterday is far more likely to buy than someone who did it 45 days ago. We build stacked audiences based on recency: * 1-7 day visitors * 8-30 day visitors * 31-90 day visitors

This allows us to bid more aggressively and show more direct offers to the most recent, highest-intent users, while using a softer brand message for those who are further out.

Third, engagement custom audiences. Don’t forget about the signals happening on Facebook and Instagram themselves. You can create audiences of people who have watched a certain percentage of your videos, engaged with your Instagram profile, or saved one of your posts. These are people who have shown interest but may not have clicked through to your site yet. They are a perfect mid-funnel audience to target with testimonials or user-generated content to build trust. A strong creative strategy is essential to make these audiences work.

Finally, strategic exclusion lists are just as important as your targeting lists. We maintain a master exclusion list for all prospecting campaigns. This typically includes all purchasers from the last 90-180 days, all email subscribers, and anyone who has visited the website more than five times in the last month. This ensures our top-of-funnel budget is spent on acquiring genuinely new customers, not repeatedly showing ads to people who have already converted or are deep in our funnel. It cleans up reporting and makes our prospecting efforts far more efficient.

The future of audience discovery for a facebook ads agency australia

The days of setting a 1% Lookalike and letting it run for six months are over. The new environment demands a more active, data-driven approach to audience management.

A continuous testing mindset is mandatory. What works today might not work in three months. We are constantly testing broad targeting against our best-performing intent-based custom audiences. We test different time windows for retargeting. We test interest-stacking against pure broad. The algorithm changes, and the only way to stay ahead is to have a structured testing framework.

This all relies on data hygiene. Your ability to build effective custom audiences is directly tied to the quality of the data you collect. This is why having a server-side tracking solution like the Conversions API is so important. As per Meta’s own documentation, CAPI creates a more reliable connection that helps the delivery system optimise your ads. Without clean, accurate data, your intent-based audiences will be just as unreliable as the old lookalikes.

We also see huge value in integrating insights across channels. An audience that performs well on Meta can inform a similar audience build on Google Ads. The search terms people use on Google can give us ideas for creative angles to test on Meta. It’s a holistic approach. Your channels shouldn’t operate in silos.

Ultimately, with targeting becoming broader, the ad creative itself does more of the heavy lifting. A compelling video or a relatable piece of user-generated content will attract the right people and repel the wrong ones. Your creative is the targeting. It calls out to your ideal customer within a broad audience.

This is the new reality of Meta advertising. It’s more complex, but it’s also more powerful when you know which levers to pull. By focusing on intent-based signals and letting the algorithm work for you, it’s still possible to achieve incredible scale and see our results speak for themselves.