Are FB Ads “Custom Audiences” dead?

When the service providers fight, the paying customers suffer.

In July 2020, Apple announced the features of their iOS14 release. Of which, a key feature is “improved” privacy protection. All developers will need to ask for permission before they can access iPhone devices on a local network.

Basically, a permission box as such would appear:


Nothing wrong, except for the negative tone it was phrased. Not surprisingly, only 5% of Apple users have allowed tracking on their devices.

p.s. iOS14 was rolled out in Sept 2020, if you have not optimised your Facebook ad manager for the changes, do look for guides on Google.

Custom Audiences are dead?!

Since the advent of iOS14, a common school of thought is that since FB can’t track their users anymore, its better for marketers to target broad but niche down on the message.

There’re many flaws to this thought beyond the scope of this article.

Instead, I share the results of my testing.

FB Ad test: Custom Audiences vs Broad Targeting

The set up

I ran a small scale test for 5 days (6 – 11 Aug), comparing the conversion rates between a custom audience and a broad audience. About $20 was allocated to each group, per day.

Variables in this test:

  • The custom audience used in this case were top-of-funnel web traffic to a client’s site, the estimated size was about 49,000.
  • The broad audience was done using interest targeting, the size was about 3M.

The objective of this campaign:

To get registrations for a webinar. A successful registration includes a name, email and phone number.


Custom audiences would convert better (i.e. more conversions, at a lower cost) given that they have had prior engagement with the client’s brand and content.

A few caveats to note:

  • This campaign was ran during the National Day long weekend were we saw a general dip in interest across all FB posts.
  • This was done in the finance niche, your mileage may differ.

How did we do?

Overall results

In the 6 days, we produced 40 registrations at an average cost of $4.91 per registration. According to our FB rep, this is below the average cost of about $10-$15 for the finance niche, which is a good sign that the ad assets were effective.

It was no surprise that the custom audience (Web Visitors) did better:

  • It generated 2x more results,
  • which were 32% cheaper than the broad targeting group
cost per conversion, by days (Custom Audience group)

On closer look, we experienced a jump in cost per conversion from 9 to 10 Aug. This coincided with the National Day celebration (9 Aug) and that many took an extra day off on 10th Aug. In the finance industry, it is not uncommon for consumers to ignore finance related content while they are on holiday.


From my tests, custom audiences are NOT dead.

1 – Custom Audiences convert better right off the bat

Instead, they tend to outperform broad audiences, especially if you’re using hotter custom audiences who already know your brand.

That said, we may need to build larger sized custom audiences, and take advantage of Lookalike Audiences to maximise FB’s current algorithm.

Consumer behavior can skew your results

In my tests, my custom audience size was about 49k, and cost had started creeping up within 3 days of running the campaign. This was probably due to the public holiday and long weekend, rather than my custom audience size.

However, there are several ways to deal with cost creep:

  • increasing custom audience sizes
  • test new ad assets – change up the copy, image, and call to action.

2 – Broad Targeting will be more expensive initially

This is expected since with this test group, we’re relying on FB’s algo to find the right audience for our campaign.

But hey, it works. Although you would have to deal with a cost per conversion that is about 50% more.

Hope for new brands

That said, new brands or advertisers can rejoice at this! This test suggests that broad targeting is still a feasible way for any new brand without an existing customer list to reach out to potential customers.

In my tests, the broad targeting audience set had not stabilised within the 5 days of testing, and the National Day long weekend did not help. If you are using broad audience targeting, give your campaign a longer runway or increase your budget.

And contrary to the belief that “learning limited” will negatively affect your ad’s performance, I have noticed that you do not need to hit 50 optimisation events in a week from your previous edits. If you’re working with a small budget, it is okay to let the ad run as long as the cost remains within your expectations.

That said, there are several assumptions taken here:

  • your product kicks ass
  • your ads are well received by your target audience (i.e. good engagement and clickthroughs)

Conclusion – We are all equally lost

Although there are FB ad gurus jumping on the opportunity to gain a few more clients and sell out FB ad courses, the truth is, we are all equally lost.

These changes are rolling in real-time, how could anyone confidently claim that they can work around it?

As marketers, we can only test to find out what works for each brand and each client.

To be honest, like all paid advertising, I seriously doubt there’s a one-size-fits-all solution to dealing with iOS 14 today.

(That said, you should be checking off the best practices such as verifying your domain, refreshing your custom audiences and more.)

Leave a Comment