Do Facebook Ads still work for e-commerce after iOS14.5?
This is a common question I get all the time. The short answer is yes, but some things have changed.
If your product or service was able to run ads with healthy returns before the update and your market landscape hasn’t changed drastically, then likely there’s an creative and audience configuration that could work for you again today, but it may take some ad set configuration testing.
Tip 1: Make sure you’ve implemented all of the instructed steps from Meta.
I won’t get into all of the technical implementation here, but I will say that it’s important you follow all of the steps to give yourself the best chance at success. This means you should make the effort to:
1. Verify your domain.
2. Setup server-side tracking.
3. Create/prioritize your standard events.
4. Monitor your Events Manager frequently and solve any issues.
If you’re running an e-commerce store, here are a few essential events to prioritize:
*Bonus: the first 4 should be quite “standard”, pun intended, but Lead may be worth considering to track if you’re collecting email sign ups on your website and/or have a higher-ticket product with a longer sales cycle.
Tip 2: Try broadening your top-of-funnel audiences.
The way top performing Facebook Ads accounts are set up now are different than before the update. The biggest change in my experience is the way audiences perform.
For years, if you spoke to most ads specialists, they’d recommend creating lookalike audiences or use quite segmented ad set breakdowns. In particular, using granular portions of lookalike audiences such as just the top 1% to 5% of the matched users. For better or for worse, in most ad accounts that I have the privilege of auditing or testing in, lookalikes are seldom the top performing audience nowadays since the update.
Over the years, Facebook has reduced many audience signals available about their users for privacy reasons and that has been accelerated by the iOS 14.5 update. Since that’s the case, I believe that the lower amount of data points on each user, along with the limited size of lookalike audiences, they’re not as accurate as they used to be and therefore the accuracy is not able to offset the decreased targeting ability of using a more narrow audience.
The acquisition algorithm of Meta Ads has improved over the years, so by allowing it to operate within a larger audience size you’re providing their machine learning the best chance of optimizing performance for you.
Tip 3: Allow your ad sets time to learn and be calculated with your changes.
The Meta Ads acquisition algorithm is arguably better than ever before, so try to leverage it as much as possible. Keep in mind that any change you make has the potential to disrupt the machine learning and temporarily reduce its efficiency.
Generally, to exit the learning phase an ad set will need to run and generate 50 result actions in a 7-day period without any major changes. Changes that could disrupt your machine learning are: increasing or decreasing your budget by more than 20% every 72 hours, changing audiences or launching/pausing ads.
Note, in some cases, even changing the budget by less than 20% can disrupt the machine learning if there are a low number of data points in the ad set or you recently made other audience or ad changes.
Tip 4: Use fewer ad sets and try to forecast optimal budget levels.
Similar to tips 2 & 3, this one also revolves around feeding the machine learning with enough data points so that you can optimize your performance. As we mentioned above, each ad set requires 50 result actions in a 7-day period which if you know what your cost per result is can help you forecast what your minimum budgets need to be.
For example, if your average cost per action result (CPA) is $15, you’ll need to spend $750 over a 7-day period to reach 50 result actions in your ad set thus making your minimum daily budget in that ad set $107.
Note, this is per ad set. If you’re running campaign budget optimization (CBO), you’ll need to factor that into your equation (e.g. if you have two ad sets in your campaign, you'll likely have to spend 2x the budget if the ad sets are receiving equal ad spend).
Tip 5: Design your creative so that it pre-qualifies your audience.
If you’ve followed the above tips, we’re now working with broader audiences and trying to leverage the matching algorithm as much as possible, therefore it’s important to think about what a user’s first reaction to your creative may be.
In a very simple example, if we’re trying to sell warm, quality winter jackets, we’ll want to show very clearly both the jacket and the words “warm”, “quality” or another variation of it in the ad image or within the first 3 seconds of a video ad.
The goal here is to train the machine learning to find users that are actively in-market to purchase a warm jacket or are at least somewhat interested in getting distracted by a jacket as they’re the most likely to actually have some intent once they reach your website.
If you're focused on building a memorable, consistent brand which I recommend you do, the fun challenge with ad design then becomes how to be creative and on-brand while designing creative that performs. I have a few tips on this, but I'll save them for another article so stay tuned for that.
It’s true, iOS 14.5 has changed the way we should run our performance-focused Meta Ads campaigns, but you can still run an effective ad program on the platform and shouldn’t lose hope.
There could be 100+ tips for me to write about here, but this newsletter isn’t the format for that, so I’ve just chosen the 5 main tips that I see most accounts struggling with these days. Stay tuned for more.
While it takes effort and thought to execute a successful paid media strategy, it can drive impactful growth for your business.
As always, I hope this article has helped you in some way. If you have any questions or would like my thoughts on your performance marketing, please don’t hesitate to reach out. I’m always happy to chat about paid media.