How Long Does It Take To 2x ROAS?

January 21, 2021

Will Hanschell

Our data of over 100 Creative AI campaigns suggests that, by working a “weekly batch” of new creative ad tests, it could be as little as 4 weeks.

We looked into 14 brands spending >$1k USD per week on Pencil’s Facebook automated ads:

8 brands were able to generate ads that beat their ROAS baseline by 80%
3 brands beat their baselines by more than 200%
Of the brands who reached the 2x milestone, half took about 12 weeks. These tended to be larger spenders who were already well optimised
The other half took less than 4 weeks. These tended to be smaller spenders with inefficiencies and quick wins on the table

Drawing a line of best fit through this data, the 100% or 2x uplift is around the 4 week mark. This is after 3-4 “weekly batches” of ads, each batch an opportunity to test and learn.

Now this data is extremely early stage and very sparse. But let’s explore why we think more datapoints might well continue to support this, and why we think 2x in 4 weeks is fundamentally possible.

From first principles, there’s a simple process almost guaranteed to improve ad performance. Almost nobody does it:

Run a weekly batch of new and different ad ideas
As soon as fatigue sets in, do more of what's working
Rinse and repeat

1. Run a weekly batch of new and different ad ideas

If you run 1 ad a week, some weeks you’ll have a winner and some weeks a dud. But if you run 10 ads a week, one will almost always work better than the others. The problem is that creating 10 ads just to throw away most of them is expensive, time consuming and painful.

The task gets even harder when you realise that to maximise your odds of finding a big winner, you need 10 ads that are totally different. Coming up with lots of different ad ideas that are all appropriate for your brand is tough, even for a professional creative with lots of time on their hands.

This is why most marketers run just a handful of new ads each month, and get stuck in ROAS plateaus. Creative AI makes it possible to generate a totally new "weekly batch" in just a few minutes.

2. As soon as fatigue sets in, do more of what’s working ‍

Now let’s say you do run your 10 ads. And 1 worked! It got a decent ROAS or CPA and scaled spend pretty well. But sure enough, it eventually fatigues and you need more ad creative. So you come up with 10 new ideas. 1 works. Pause the others and scale your winner. If you can keep this up, week in, week out, you’re managing ad fatigue and doing better than many, many other performance marketers.

The problem is, you’re spending a ton of time, effort and money to stay in the same place. You’re avoiding fatigue, but you’re not really learning anything from what you’re doing or making consistent progress.

3. Rinse & repeat ‍

This is where Creative AI comes in. Generating new ads frequently enough to avoid fatigue is difficult even for dedicated creative teams. And even the most savvy creative teams find it difficult to accurately learn from what works.

Because every AI-generated ad has been generated from individual known elements (messaging, tagged assets, branding) it can be deeply analysed. Last week’s winners all had prominent branding? Product in the first 3 seconds? Used a specific USP? These are insights you can build on, and they’re automatically extracted from every piece of AI-generated ad you run.

But even this isn’t enough. Running the same ads all the time creates an echo chamber and doesn’t fully address fatigue.

You need to go beyond what’s working and wade into what might work. Creative AI's predictive insights allows you to generate ads similar to past winners, but which also contain new ideas that might just work even better.