From ad testing to ad scaling: when and how to make the leap

Learn when it’s right to move on from testing and onto one of three proven methods for scaling winning ads.
Sam Russo

People often get caught on the ad testing treadmill. They test, they learn, they test, they learn — but they struggle with understanding when to take those learnings and actually roll them into ROI and results.

That’s what we’re about to break down for you in this article:

  1. When to keep leaning into testing your ads on Marpipe
  2. When and how to move on from testing and scale what’s working

When to keep testing your ads

If the ads you tested weren’t able to reach a high level of confidence, you’ll want to keep testing.

Just because you didn’t find any high performing ads doesn’t mean you didn’t learn something. Believe it or not, data from losing ads can be just as valuable as data from winning ads. Dig into the creative intel from your test to help you understand what to test next

Most importantly, stay the course and keep the faith. A common question we hear is, “if my ads didn’t test well, what did I do wrong?" Creative testing is not something to be done sporadically or short term. It’s something that pays off over time, as you continue to iterate, learn, and follow the data.

All of this isn’t to say a single test can't create a winner. It’s possible to find at least one ad that is performing strongly compared to your goal, especially if you start your testing based on ads that have worked in the past. 

And this is where we can start to scale.

When to scale your ads

Any time you find ads that performed close to or better than your KPI goals, you should consider scaling them. There are three main ways we see to scale your top performers:

Scale winning ads directly from your test campaign

This may be the best way to scale up winning ads while keeping their performance as strong as what you saw in the test. 

To understand why this is the best option, you have to understand how Meta treats new ads introduced to its platform.

The moment a new ad is launched on Meta, it enters what Meta calls the “learning phase.” The learning phase is a period when Meta is determining the best way to deliver your ads and ad sets for optimal KPI performance. An ad only exits learning phase once it reaches 50 conversions.

When running your test in Marpipe, your ads and ad sets have already spent some time in Meta’s learning phase. So scaling directly from your test allows your ads to keep the learnings they have gathered and scale quicker. Meta has already learned about the ad/ ad set combination and knows how to find conversions for your chosen KPI.

To scale from your test campaign, all you need to do is pause poor performing ads/ad sets and increase spend on the top performers. We typically suggest scaling at least a couple of top-performing ads to start, instead of only scaling just the #1 performing ad. For example, if you ran a 30-ad test, consider scaling the top 5 to 10, pausing the low performers of the group after a few days, continuing the remaining winners a week and optimize from there.

There is one inherent challenge with this scaling option. Because you now have scaling campaigns that were previously test campaigns, your ad manager account structure can start getting messy and confusing. It’s possible to avoid these issues if you follow a systematic approach to naming ads, ad sets, and campaigns. 

Scale winning ads in a new or evergreen campaign

Many advertisers like to duplicate and move winning ads to their separate scale or evergreen campaigns that live within Meta. This is great for keeping the ad manager account organized and manageable, for keeping audiences consolidated to avoid too much overlap, and for letting your evergreen winners act as somewhat of a baseline for new test winners to go up against.

However, when you duplicate a winning ad, that ad/ad set will have to go through learning phase all over again. This, of course, can have an impact on your ad’s performance.

Oftentimes, advertisers will write off new ads that just entered an evergreen campaign too quickly because they do not outperform the long-term winners right away — even though the test environment made it look like they would have. Test environments are not perfect indicators that an ad will scale immediately in any campaign. They are simply meant to guide you in the right direction creatively while helping you find new ads to try scaling. 

This is all to say that you should give new winners from your tests a decent amount of spend at scale before writing them off in an evergreen campaign. We also suggest scaling a handful (5–10) of top-performing ads with this method as well. If a winning ad from a test doesn’t scale up above a long-term winner once you’ve given it ample time to success, that’s a good sign that more testing on that ad concept is in order. On the flip side, the creative elements that make up your winning evergreen ad would give you a good indication of how to test further.

Scale winning ads in both test and evergreen campaigns

Upon finding winning ads on Marpipe, you can scale them from within both your test campaign and also your evergreen campaign.

This is the most testing-centric option of the three because it gives your test winners different environments in which to prove themselves. Said differently, you’re A/B testing winners against themselves: in a campaign where they have already gone through the learning phase and in a campaign where they have not. After some spend is deployed evenly to each, you can optimize, likely having found a few strong performers in both campaigns that you can continue running as long as they perform against your KPI goals.

The challenge of this method is that you’re essentially scaling times two. This means you’ll need to spend more budget to see which approach works best than you would if you were to scale using just one method. To avoid spending too much on this "A/B test" after having just run a test campaign, you can start your optimization process a bit sooner than you would with the other scaling techniques.

Two sides of a business-critical coin

Testing and scaling go hand in hand, especially when you can find the right balance between them and understand how to strengthen one with the other.

Found a winning ad through testing? Great! Scale it while you test it against even more variants. Have an evergreen ad that used to perform well but has started to decline? No worries. Swap in some new creative elements to see if you can give it a boost, then scale the new winner.

The more you test and scale, the greater your ability to do both strategically will become — a mastery your brand and business will undoubtedly reap the rewards of.

Not sure which path to take? Download our Test-Scale Decision Tree to make the right call.
Boost ad performance in days with a 7 day free trial.
Claim Trial

How to Run a Multivariate Test

The Beginner's Guide

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Tiffany Johnson Headshot

How to Run a Multivariate Test
The Beginner's Guide

Plus, Get our Weekly
Experimentation newsletter!

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Outliers is the weekly newsletter that over 10,000 marketers rely on to get new data and tactics about creative testing.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Sign Up to Outliers 💌

Join 7,000+ marketers in the world’s best newsletter all about catalog ads!
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.