Back
All Resources

How to Improve Your Top-Performing Creative in 3 Easy Steps

Using MVT to Make Incremental Improvements to Your Most Effective Ads
Brett Friedman

You’ve got an ad that absolutely crushes it month after month, year after year. You’ve done everything you can to replicate it, to match it, to keep its juice from running low. And for the most part, it still does alright. But what if I looked you in the eye and said you can do better

Would you believe me? You should. Because you can always do better. Stretch that comfort zone thin. Test your boundaries. Test your ads. 

If you want to make your top creative better, follow this 3 step process:

  1. Tag your top-performing ads' creative components
  2. Make creative assets to replace those components
  3. Multivariate test your ads to find component-level improvements

Step 1: Tagging your top-performing ads' creative components 

Let's say your top-performing ad looks something like this:

I know it says it's not an ad, but it is. A damn good one.

There are tons of creative components here. In fact, almost anything can be a component. Here are most of the possible components we could tag:

Nine Components

To start, you should tag every component you think you can consider a component. This is a simple ad, yet there are nine components. Your top-performer may be more complex. That's okay, but get as granular as you can. Tiny changes often have large impacts on performance. For example, changing the text alignment from center to top right in this ad increased CPR 215.38% and making the text color blue on green increased CPR 130.76%.

Changing the text alignment from center to top right in this ad increased CPR 215.38%

Those are the largest single component differences from our test, but it speaks to the amount of variability possible in even your award-winning ads. To clarify, tagging just means identifying differential characteristics in your ad. Typically, they're individual creative assets, but not always (as in the case of layouts and designs)

Step 2: Make creative assets to replace those components

Now that you know your components, you need to make a couple choices.

  1. What components are likely affecting performance the most?
  2. What changes can I make to those components?

The components you choose to vary are both extremely important and completely arbitrary. Anything you systematically change, you'll eventually find big differences.

On the other hand, some component changes have much more impact on performance than others. So just choose what you think people like or dislike the most about your ad. It might be the background color, it might be the layout. Just choose.

We chose to start with a copy test because we felt the copy drove this ad's performance the most given its uniqueness. Here are the components we used:

Three Components

That's only a third of the possible components listed earlier, but it's perfect to start. Especially since we need to choose the values of those components. To be clear, when I say values, I mean variants of a component. Not company or human values. Here are the values for this particular test:

Experiment Variable Outline

Again, you might think product or image or background are more important in this ad. The nice thing is, you can test those later. Just choose what you think will affect performance the most.

Now comes the fun part. For each of these values, you're going to make a new creative asset or in the case of text alignment, a new ad. Here are our four text color assets:

Four Text Color Values

our two font types:

Two Font Type Values

and our nine text alignments:

Nine Text Alignment Values

Now, we do the real work.

Step 3: Multivariate test your ads to find component-level improvements

Once you know the components you're testing and you have creative assets to vary them, you need to put them together in a multivariate matrix. So you start with your original, top-performing ad in its full form. You choose one of your components. Then you switch the original assets for the new assets one at a time.

Starting with text colors, we trade the black text on white background for white text on black background, keeping everything else the same. Next, we'll switch text color to white on transparent, then colored on colored so we have every iteration of text color.

Here's where it gets tricky. You move to your next component. For us, it's font types. We already have four text color variants. Now we'll switch the original sans serif font type to serif. Combining that with each text color, we end up with eight variants:

  • Black text on white background x Serif
  • White text on black background x Serif
  • White text on transparent background x Serif
  • Colored text on colored background x Serif
  • Black text on white background x Sans Serif
  • White text on black background x Sans Serif
  • White text on transparent background x Sans Serif
  • Colored text on colored background x Sans Serif

Let's take it a step further!

To spare your screen, I won't type this one out. But when we change text alignments, we get seventy-two variants. Each of those eight variants above needs to cross with each of the nine text alignments. That's where you get your multivariate matrix from just one ad.

Check it out:

Seventy-Two Ad Variants

Now all you need to do is run them live on your favorite ad network! That's where you'll finally see your top ad of all time get even better. Maybe it's your background color, maybe it's your copy, maybe it's your model. Whatever it is, you'll find it and you'll know. To make sure you know, make sure you name your ads according to the components in them. That way you can isolate components and values that drive performance, not just the best ads.

Anyway, not only does this method give you ads that perform better than your top ad of all time, it also gives you leeway to continue improving it over time. If ad fatigue exists, it's no match for your new methods!

P.S. after testing copy, we tested background color to find a 117.56% performance increase and product shots to find a 14% performance increase.

For another interesting MVT use-case, see our article on product-market validation.

Happy testing!

Read our full creative testing guide

SIGN UP FOR THE EXPERIMENTAL METHOD NEWSLETTER
TO GET THE GUIDE

Download Now
View Sources

Heading 1

Heading 2

Heading 3

Heading 4

Heading 5
Heading 6
"This is a block quote"
  1. This
  2. is
  3. an
  4. ordered
  5. list

This text is bold

The rich text element allows you to create and format headings, paragraphs, blockquotes, images, and video all in one place instead of having to add and format them individually. Just double-click and easily create content.

A rich text element can be used with static or dynamic content. For static content, just drop it into any page and begin editing. For dynamic content, add a rich text field to any collection and then connect a rich text element to that field in the settings panel. Voila!

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.