How time-consuming have your Facebook campaigns become? Are you struggling to manage campaigns with ever-growing numbers of ad sets and ads? Good news! Bringing automation to your Facebook ads could give you some of your valuable time back, while still providing the results you need.
Rather listen than read? Check out the latest episode of Test. Learn. Grow., as Media Manager Keefer Kopco talks about how we’ve leveraged some of Facebook’s automation tools to streamline our campaigns.
Innovation and Audience Expansion
Historically, a key benefit of digital marketing has been the ability to target very specific audiences; the narrower that target is, the higher the quality of the audience. But, over the last couple of years, there’s been a shift in advice from our Facebook representatives: use larger audiences and use objectives for our campaigns that focus on automation.
Expanding audiences out of these narrow segments enables machine learning within the platform. The algorithms that have been designed are able to make decisions that humans can’t make quickly enough, and they’re able to look at the audience on a macro level. Machine learning determines the chunk of people most likely to convert, but the audience must be large enough to allow for this.
The Streamlined Campaign Structure
Because we were previously focused on zooming in on the narrowest of audiences, we were faced with some situations that included a half dozen campaigns, each campaign housing three to five ad sets, each ad set a target. Conversations evolved with our Facebook reps from “enlarge your audience” to include “limit both your campaigns and your ad sets.” The goal now was to group as many people together as possible—consolidate your audiences. This gives the algorithm a chance to look at things on a much larger level and determine who will convert.
The Testing Begins
At the beginning of 2021, we rolled out our streamline test: one client, one campaign with a key objective. This consisted of one extremely large audience at the ad set level and then several different ads at the ad level. We wanted to explore different creative and see what kind of person responded to each. The result? It didn’t outperform what we’d already been running.
Where was the failure point? Test, learn, grow is a way of life at Level Agency, so for round two, we weren’t quite as aggressive. For two different clients, we used two ad sets instead of one and found that, after two weeks, our results indicated we were on to something. Over the course of 30 days, we saw that this test dropped the cost per conversion by as much as 50%, and that stayed fairly consistent. To this day, those two clients have maintained those lower costs per conversion. Eventually, we were able to go back to the client from the initial test and present our new findings and try again—they had the same success we had with the other two clients: a more than 50% drop in cost per conversion as soon as we were out of the first Facebook learning cycle, and they’re maintaining that lower cost.
Broad Targeting
Where do we go from here? Is there an opportunity to streamline further and reduce costs even more? Our latest conversations with our Facebook reps have centered around Broad Targeting—targeting based on location and age with exclusions added. For example, exclude anyone who’s previously converted and anyone who’s currently in your sales funnel, effectively targeting by negation. We’re finding that this streamlined form of targeting is an even bigger runaway success. We’ve seen as much as a 70% drop from the initial cost per conversion.
For our next steps, we need to continue making those audiences bigger, expanding them as much as possible to iterate the process. This includes dynamic creative, too—giving Facebook one ad with different copy and image options and letting them determine which combination works. The more autonomy we give to the automation tool, allowing it to make the decisions humans can’t on a day-to-day basis, hopefully continues to give us better results as it understands what success looks like.
The Human Element
Feeling skeptical? We understand the concern—no one wants to lose the human element. Our goal is to demystify Facebook Automation a bit, but we always recommend running parallel tests with more traditional efforts. The biggest point to make when people are talking about automation is that humans are still making decisions at the end of the day. We’re just trying to create structures that allow automation to view things at a much more refined level. And ultimately, automation can’t create the campaign, set itself up for success, or report out the results.
What did we miss? Give us your thoughts in a comment!
