Facebook’s Significant Edits: How to Minimize the Risk of a 20-40% Hit to Your ROAS
Don’t mess with the algorithm.
That’s the fundamental idea behind Facebook’s significant edits. There’s a lot more to it, of course, and we’ll get into all that, but here’s the primary reason why you don’t want to make a change that Facebook construes as a significant edit: It will move your campaigns into “the learning phase,” which in turn will suppress your campaigns’ ROAS by 20-40%.
The learning phase isn’t meant to be some kind of punishment. It’s built to optimize your campaigns. But once it’s triggered, you’ll probably see a drop in your ads’ performance until the system has accrued enough data to shift your campaigns or ad sets out of the learning phase and into full delivery. Usually, getting out of the learning phase requires at least 50 conversions within a week.
What is the “learning phase,” exactly? Here’s how Facebook describes it:
The learning phase is the period when the delivery system still has a lot to learn about an ad set. During the learning phase, the delivery system is exploring the best way to deliver your ad set, so performance is less stable and cost-per-action (CPA) is usually worse. The learning phase occurs when you create a new ad or ad set or make a significant edit to an existing one.
And here’s what it looks like in your campaigns:
Clearly, slashing 20-40% of ROAS is not a good thing. And unfortunately, an advertiser who doesn’t understand the principles behind significant edits could erase the benefits of any improvements they had been making with their changes.
Here’s an example of how that could happen:
- Say an advertiser makes five changes to their campaign over the course of 30 days.
- Each of these changes triggers a significant edit and moves the campaign into the learning phase for three days.
- While in the learning phase, the campaign loses 30% of ROAS.
- Those five significant edits have forced the campaign into the learning phase for a total of 15 days out of the 30-day period.
- Over the course of those 30 days, the campaign has lost 15% of its total ROAS.
- The five changes the advertiser made will have to generate more than a 15% improvement in ROAS in order to overcome the losses incurred by being in the learning phase.
Now, does that mean it’s never a good idea to go in and make a change that could possibly trigger a significant edit? No.
But it is smart to make strategic and limited changes to your campaigns. For example, don’t just try one bid strategy for a week and then switch back. Give the algorithm enough time to gather the data it needs to optimize your campaigns.
Another well-tested trick is to just leave well enough alone. Instead of changing an ad set or campaign, duplicate it, make the change(s) you want, and then see how well your optimization works. This preserves your campaign, but it is not an ideal solution, as duplicating a campaign or ad set means:
- You are now spending more money.
- You’ve probably just created an audience overlap.
- Your new campaign or ad set will also have to go through the learning phase.
But if you’ve got a campaign or ad set that you really want to change, but you really don’t want to disrupt its current performance, duplicating it may be your best bet.
Of course, you can always just go ahead and make the change to the existing ad set or campaign. But before you do, at least understand which changes are most likely to trigger a significant edit.
How Significant Edits Get Triggered
There are four guaranteed ways to trigger a significant edit. Any changes to these campaign or ad set elements will do it:
- Bid strategy
- Optimization events
- Ad creative (including adding a new ad to an ad set)
Also, pausing an ad set or a campaign for seven days or longer will also be interpreted as a significant edit and will shift the ad set or campaign into the learning phase once the ad set or campaign is restarted.
These are changes that may or may not be interpreted as a significant edit, depending on how large the changes are:
- Switching your campaign bid strategy. This may cause multiple ad sets within the campaign to reenter the learning phase.
- Changing the spending limit amount of an ad set.
- Changing bid control, cost control, or ROAS control amounts.
- Changing an ad set or a campaign’s budget amount, unless you’re using the target cost bid strategy within Campaign Budget Optimization. Then any budget changes aren’t considered significant edits.
So what’s enough to be interpreted as a significant edit, and what’s not? Is a 21% increase to a campaign’s budget okay, but 22% is not?
Sorry, but there’s no way to know for sure. There isn’t a specific threshold for what will trigger a significant edit. But I can tell you for sure that a 70% change to a budget is more likely to flag a change as a significant edit than a 20% change is. Generally, a rule of thumb is that changes of over 30% are at risk of being categorized as significant edits.
Note that it’s also possible to make a major change at the campaign level (say, to campaign budget) and have some, but not all, of the ad sets within that campaign be shifted back to the learning phase as a result of the edit.
Making two changes to a campaign at once are also more likely to have Facebook mark your work as a significant edit.
How to See Which Changes Have Triggered a Significant Edit in the Past
Concerned about how much performance you’ve lost due to significant edits and the dreaded learning phase that follows them?
You can see which changes have been designated as significant edits in your account by using the Inspect Tool at the ad set level and looking at the well-named “Significant Edit History” report.
On the Inspect Tool page, scroll down to the bottom where the “Significant Edit History” report is:
Significant Edits are Just Part of Increasingly Automated Facebook Advertising
If you’ve been paying close attention to the announcements from Facebook over the last few years, the principle behind significant edits should be familiar. The algorithm is increasingly taking over tasks that human UA managers used to do. And that’s a good thing.
So per the Structure For Scale best practices Facebook released late last year, the more data you can give the algorithm to work with and the less you restrict its learning, the better your campaigns will run. Instead of crunching numbers and trying to squeeze every drop of quantitative optimization out of your campaigns, we recommend you focus on creative and creative strategy.
Brian Bowman is CEO of ConsumerAcquisition.com.