How To A/B Test Ads For Your Most Successful Campaigns

When you create your marketing campaigns, and are working on your creatives, the need to just go by gut feeling can be strong. But why base your marketing decisions on gut feelings when there’s a surefire way to get data to show you the way. We’re talking about A/B testing. By learning how to A/B test ads you’ll be much more successful with your campaigns.    

This technique will help you validate, or invalidate, your theories and enhance your customer conversion processes. And with it on your side, you’ll be able to get a better ROI on your marketing and approach campaigns with confidence. 

What exactly is an A/B test?

Before we get to how to A/B test ads, let’s take a closer look at what this means. A/B testing refers to using two versions of any given marketing asset to identify which generates the best results with an audience. It’s important to note here that the difference in the two versions should not be drastic so that you can assess what is making an asset a hit.  

On an average day consumers come across a wide variety of marketing creatives that are all trying to get them to take a particular action. These include, but are not limited to, display ads, promotional emails, campaign landing pages, onboarding flows and various forms. 

When you look at how to A/B test ads that you’re running you can identify exactly what to tweak to get your conversion rates as high as possible. This could mean running the same ad, but with a longer and shorter version of the copy. Or using the same design, with the exception of different background images.  

A/B testing works by dividing a brand’s audience into two groups. Then each group is targeted with a different version of a marketing asset. The two groups will be…you guessed it, “A” and “B”. A is the control group and B is the treatment group. This means that group A will be shown one version of your creative and group B will be shown a different version of it.

A note on multivariate testing

As you research how to A/B test ads you’ll likely come across multivariate testing. And you’ll probably get very excited about it since it includes testing for multiple variants AND testing for multiple elements as well. The end goal with multivariate testing is to determine which particular combination ends up performing the best.

The thing is you’ll need to have a ton of traffic to be able to conduct multivariate testing in a productive and helpful way. So for the time being file that away as something to come back to. 

blank

What are the benefits of A/B testing? 

There are many different types of split tests that you can run to see what gives your brand the best returns. Whichever type you opt for, the common goals and benefits that comes with A/B testing are: 

  • Increased website traffic. Testing blog posts and webpage titles can change the number of people who click on a link. Figuring this out can generate more traffic for your website.
  • You can increase conversion rates by testing different colors or locations for your CTA buttons, or even anchor text on the call to action buttons. These changes can increase the number of people who subscribe, submit a form or convert in any given way.
  • It can lower the bounce rate. If you have realized that a lot of people leave your web page or that a specific ad is not getting enough conversions, testing with different fonts and colors or copy can help you reduce the number of people who are bouncing off your site or ad.
  • About 40%-75% of ecommerce shoppers leave a site with items in their shopping cart. You can lower cart abandonment rates by testing different fonts, images and colors, or even placement of CTAs.

Now let’s take a look at how to A/B test ads, along with tips and best practices. These will help you get going, even if this is the first time you are trying out A/B testing. And if you need a hand getting all of your designs ready to A/B test, get yourself a Kimp team and get all the designs you need for a flat monthly fee.

blank
How to A/B test ads: Identify your goals for A/B testing

You will be able to monitor a number of metrics for each test you run. Which is great but keeping a close eye on all of them isn’t effective for the purpose of A/B testing. Choose one key metric that you want to focus on. 

This metric, aka your dependent variable, should be determined before you run the test. And ideally it should be decided even before you set up your second variation. 

Where do you want this variable to be at the end of the test? Knowing this will help you make decisions in favour of your end goal. And it’ll help you optimize every aspect of your A/B testing. If you wait until after your test to determine the metrics that are most important you might find that your testing is requiring extra steps and more time than you’d like.

How to A/B test ads: Start with picking one variable for the test 

You may feel like there are many different variables that you want to test, but it can be very difficult to do that and get reliable results. Especially when you’re starting out with testing your ads. So get started with just one independent variable, at any time, and measure its performance. Otherwise, you’ll have a very hard time figuring out which variable is causing changes to the metrics. 

Take a look at the various elements in your ad like the design, layout, wording and colors used. You can test for any of these. If you are running an email campaign, you can test for the subject line and the ways in which you can personalize your email. Remember that even the most simple changes can make for significant changes. 

The simpler the changes you make, the easier it will be to measure the results and performance.

How to A/B test ads: Identify your control and your challenger 

Once you have determined your dependent and independent variables, as well as your goal, you can use that information to set up the control. The control is the current version of the ad campaign that you’re running or a design that uses the same style, elements and copy as usual. 

Then, start creating the variation or the challenger. This is what you will test the control against. For example, in your control ad campaign, you may have a call to action that says Buy Now. In your challenger, you can include a call to action that says Shop Now.

How to A/B test ads: Split the audience 

If this is the first time that you are running an A/B test, the first challenge is to randomly split the audience. It is very important that this split is random. If you’re using a platform, as you would for email campaigns, check to see if it has a built-in-tool for this purpose. If not, give Unbounce, Optimizely or Google Optimize a try. 

As a general rule of thumb, the two groups should be as similar as possible. If you divided your audience by gender and then ran the test for instance, you will get a result that is affected more by the audience than the actual variable. 

Of course for some tests you won’t have a set audience that you’re splitting from the get go. Keep reading to find out how to navigate that scenario.

blank
How to A/B test ads: Consider sample size

You should also decide on the size of the sample audience based on the testing tool and the kind of test that you want to run. Some tools will require that you have an audience of a particular size.  

When it comes to the kind of test, some will allow you to have more control over the audience than clothes. For instance, if you are running a test on emails, you have a set list. And you can send out the test to just a small portion of your email list to start. This will allow you to get more significant results. Once you pick the winning variant, you can then send it out to the entire list. 

In the case of testing marketing assets that don’t have a set audience, like a landing page, the amount of time that you run your test will determine the size of your audience. And that will determine your results.

So be sure to test for long enough that you can see a significant difference in the statistics between the two control and challenger.

How to A/B test ads: Decide how notable results should be 

Next it’s important to think about how notable your results need to be in order for you to choose the control over the challenger. Or vice versa. The importance of the statistical significance of A/B testing shouldn’t be underestimated. Ultimately, the higher the percentage of confidence the better results you will have. 

You may want to be as much as 95% confident before moving forward, if the experiment took a lot of time to set up. 

At the same time there’s a few other things to consider. For instance, when you are testing for something that is highly specific (like a button color), but only subtly impacts conversion rate, you’ll want to make sure you are being more methodical. This is because you can actually take a very precise approach here by modifying one element at a time.

But when it comes to more drastic changes, that could potentially increase your conversion rates by a lot, you may be more willing to forego statistical significance. This occurs in the case where you want to try a completely new approach or design style because your current creatives are just not working. 

It really comes down to knowing what odds you’re comfortable working with. Some things are worth taking bigger leaps and risks for.

How to A/B test ads: Run only one test per campaign at a time 

If you test for more than one thing in a single ad campaign, you are setting yourself up for complicated results that will be hard to decipher. For example if you are running an email test that directs people to your website, don’t run another test directing to your landing page as well. You won’t know which test caused which results. And in turn you’ll have less clarity about your leads to work with.

blank
How to A/B test ads: Test at a granular level (for the most part)

One of the most common errors that most marketers make is testing two variants that are extremely different from each other. If the control and challenger creatives are too different from each other, you may not have any actionable data. 

When there is a lot of difference between two versions, it will be difficult to see which aspect caused an increase or decrease in conversions. There is also a misconception that all variations in the test have to be dramatic and massive transformations. Don’t buy into that hype if you’ve already got an ad or landing page that’s getting you some results. In this case your goal is just to increase those results.

A simple change like colors, fonts or copy as we said before, can be subtle but cause a big difference. Sometimes, all it takes is using a different call to action that will drive people to engage. In fact, punctuation can change the behavior of customers as well. 

But there is an exception. Maybe you are dealing with a marketing asset that you haven’t had much, if any experience with. Or want to go in a completely new direction with a creative. In this case using two very different concepts has merit, early in your testing process. Based on how they perform you can make one of them the control in a new test. And keep tweaking from there.

How to A/B test ads: What should you do with the results? 

Most tools that you use as part of the A/B test process will allow you to track the progress even while the test is running. This will help you see which variation is showing a better performance. If the improvement is a very obvious one, across a substantial enough group, you may just be able to end the test early. 

But what would you do if that is not the case? Let’s assume that you are running an A/B test for a banner ad. Based on the weekly traffic that you usually have, you were expecting an increase in 25% by the end of 6 weeks. However, 6 weeks have passed and you are only seeing a 10% increase. 

Since it is difficult to have confidence to act on a small change, you would technically have to keep the test running for 32 weeks at the same pace to get an audience sample that is large enough. This will then allow you to have confidence in that 10% increase. 

Unless this banner ad is very time-sensitive, and that is just too long of a duration to wait. In most cases what marketers do in this scenario is identify that the change is too small for them to have confidence. They then focus on shifting gears to run another experiment that may provide them further clarity. Essentially, it is important to know when to allocate time and energy to a test and when not to. If there is a chance that you can invest that time and energy in a different test that will yield better results, go for it.

blank
How to A/B test ads: Test, tweak, repeat

Any improvement that you get from A/B testing is iterative. This means that with each test you run, you will learn more about your customers and what motivates them to buy from you. Similarly, it can also tell you what drives customers away from your brand. 

As you get the hang of running these tests, you will be able to form better hypotheses and also figure out more impactful tests that you can run. This will eventually help you get a broader and more satisfied customer base. The key here is to keep on testing. In any business, the culture of continuous improvement is one that will help you grow and thrive. And A/B testing can help you foster this culture within your brand too.

How to A/B test ads: Ask for feedback from actual users 

A/B testing has a lot to do with quantitative data, and it’s helpful in many different ways as we’ve mentioned. But it does lack in one important area. 

Figuring out how to A/B test ads won’t help you understand why people take certain actions over others. Just that they do. So we suggest that you also collect some qualitative data while you are at it. 

Basically what you need to do is collect feedback from your actual users. You could easily conduct a survey or a poll that will get you this qualitative insight. For example, adding an exit survey on your site where you ask people why they choose not to click on a certain CTA. Or you could have a survey on your thank you page through which you ask customers why they clicked on a button or submitted a form. 

You may find that sometimes while people click on a link that takes them to an ebook, they do not convert when they see the price. That is the kind of information that will let you know the reasons motivating your customers’ behavior.

Don’t dismiss the importance of A/B testing

While it can certainly feel like everything from daunting to mundane and tedious, don’t put off testing your ads and marketing creatives until the last minute. Or until you’ve poured tons of time, energy and money into your ads. If you start running tests earlier on in a process, you will have actual data about what customers do to help you run more successful campaigns. And you will not be making decisions based on what you think they might do.

This will help you create better ads and landing pages. And in turn, your campaigns will give you better leads and conversions. Don’t be discouraged by all the details and technical terminology. There are many resources and tools out there that can help make the process easier for you. 

Useful resources for your A/B testing 

Here are some useful resources that will help you figure out how to A/B test ads for your campaigns. Try going through them before you get started, and come back to them anytime you need more tips:.