It's Time to Revisit Creative Testing

So this is a topic I’ve been thinking about a lot lately, and honestly, I’ve been struggling with how we test creative for much of the past couple of years. But I couldn’t put my finger on what bothered me so much.

I know that I disagree with the focus advertisers put on creative testing these days. I also know that I don’t do many of the things other advertisers do. It’s felt like a waste of time and resources. And I think I’m finally able to clearly articulate why.

I’ve been doing this for a really long time, and I admit that’s not always an advantage. It can be a disadvantage when you’re stuck in your ways or resist rapid change. There’s a get off my lawn element that I’ve had to work hard to avoid.

But the advantage is perspective.

Not only have I seen so much over the past 14-plus years, but I’ve also done it. And most importantly, I’ve documented it.

I’ve referred back to my blog so many times to confirm when something changed or how things used to work. And how we treated ad creative is a big one.

Things were so much simpler back in the day.

I went back to my old blog posts to confirm this, and the screenshots were just... crazy.

Prior to 2017, one ad was a very basic construction. It consisted of:

• One image or video

• One primary text

• One headline

• One description

We couldn’t even customize by placement yet. And you may not even realize how simple the placements once were.

Back in 2013 or 2014, it was just Facebook desktop, mobile, and right column. That’s it. Each ad was a single combination of ad copy and creative. One combination.

Creative testing back then meant creating multiple ads to see what worked. You could actually isolate the best-performing combination.

But things started to change at the end of 2017.

By then, there were 12 placements. And Facebook introduced Dynamic Creative, which allowed us to provide:

• Up to 10 images or videos

• 5 primary text options

• 5 headlines

• 5 descriptions

All for a single ad.

That changed a lot. Dynamic Creative was really the precursor of today’s ad creation. Much of that is still around in some form.

Then in 2019 — still six years ago — we were finally able to customize creative by placement. Before that, if you wanted different creative by placement, you had to create a separate ad set and ad.

So already, things were much different. One ad could now have many variations.

And look at where we are today.

There are 26 different placements. We can customize both text and creative for all of them.

Meta asks us to submit three aspect ratios, but we could fine-tune it even more. The five primary text options and five headlines are baked into standard ad creation. And now you can even add five more AI-generated versions of each.

So in the past, there was just one possible combination. Now, thanks to variables like Advantage Plus creative enhancements, site links, AI-generated text and images, and more — there are hundreds or even thousands of potential variations for a single ad.

Back then, it made sense to isolate the best-performing combination and discard the rest. Now, that's impossible. And unnecessary.

Ads work differently now.

A different version of your ad will appear to different people depending on placement, device, content consumption habits, and what Meta believes they’ll respond to.

Many combinations aren’t just okay — they’re essential.

A high-performing ad today will include many high-performing combinations of copy and creative.

Here’s a simple example. You’ll get different engagement from a 9x16 video on Instagram Reels than you will from a square image in the Facebook desktop right column.

You’ll probably get more engagement from the Reels video. But does that mean the video version is better?

Should you remove the static image? Of course not.

That video wouldn’t even work in the right column. You need the best option for each placement.

Now consider how people consume content across 26 placements. Think about how every person is different. The idea that there’s one best combination doesn’t hold up.

This doesn’t mean you shouldn’t learn what works. Monitor which images, videos, text, and headlines perform best. Apply what you learn to future ads.

But the days of identifying a single winner and discarding the rest are gone.

Here’s the bottom of the glass.

Question whether your current method of creative testing is still helpful.

If you’re building separate campaigns or ad sets to isolate combinations, ask yourself why. What are you trying to prove? Could this be counterproductive?

Auction overlap is real. And the results from testing may not carry over anyway.

My approach is to keep it simple.

When I launch a new campaign, I include all the ads I plan to use. If it performs well, I leave it alone. If not, I think about new angles and create additional ads. And yes, I add them to the same campaign and ad set.

It will restart learning. That’s fine. Sometimes that’s even helpful.

Whether you follow this same approach is up to you. But here’s the key takeaway:

You no longer need to find one best combination of ad copy and creative.

You need many high-performing combinations to get the results you want.

And for more on how this evolved, check out my recent blog post at jonloomer.com/control.