Back arrow

Improve your campaigns with A/B Testing

Improve your campaigns with A/B Testing


Dear media buyers, 

I hope you made good money during the last weeks! Halloween & Black Friday are now behind, but Christmas and New Year are close, so there’s no rest for us as digital marketers! 

I’m sure that not all campaigns you have started are profitable.. Am I right? I know it’s the reality for every media buyer, and it’s normal! 

Luckily for us, we have various tools and instruments that help us when it’s time to test different approaches and campaigns. One of these instruments is the A/B test (a.k.a split test) and today we’ll review it in detail.

So, what is the A/B test exactly? 

It’s the method of comparing the performance of two versions of landing page, angle, ad copy, creatives, email template, etc. In media buying, we usually do this at the start of our campaigns, when we have so many doubts and unknown information. 

And, how is this done? 

It depends on what we want to test. If we’re testing our creatives (images, ad copies) we’ll do it on the traffic source level and, of course, we’ll continue only with the creatives that are profitable (or close to profitable) with a good EPC and CTR. 

If the goal is testing landing pages, then you can easily do that on your tracker - just distribute the traffic equally between the landing pages and keep the best performing one(s). 

No doubt you’ll also know that split testing the campaigns is extremely important too and can be done in the same way - send the traffic equally to the offers in your tracker, and after gathering enough data, stop the unprofitable ones. The most significant metric here is the EPC - the higher, the better (always send more traffic to the offer with the best EPC!). 

Don’t forget to split test your angles too - they are crucial for your campaigns. 

How can you test them? 

Start with the creatives - use 1 angle with two different images - and continue with the landing pages, transfer the angle to them. 

Only after you have winners in these elements, you can A/B test your CTA buttons, colors, etc. 

I know some of you will say this is not that important, but let me explain you why split testing is so important in this industry! 

1. When you do it correctly, A/B testing will increase the conversion rates and that means more money in your pockets. Pretty good reason, right! 

2. It will help you understanding your audience, what they want and what their needs are

3. If you practice split testing regularly, you’ll have more knowledge of what works right now and therefore you’ll have a competitive advantage against your competitors

That’s it with the theory! 

Let’s continue with some practice - I’ll show you some creatives and explain why they’re good or bad for split testing purposes. 

The first vertical is my favorite - dating! 

At first sight, both creatives are good, but this media buyer can improve the split testing by changing some elements: 

1. Use the same image for both creatives and different angles (text)

2. Use different images and the same ad copy 

The idea of testing a creative without a CTA button against one that includes a CTA button is good, and it’s also a good practice to test emojis. I would say good job :)

Another set of creatives. Do you agree with me that this media buyer’s job regarding split testing was not the best? 

Testing different topics is a good idea, but at least one angle has to be the same. Here, there’s no common angle. Messages are different, pictures are different, etc. Something has to appear in common between the two creatives in order to have a reference of what works and what doesn’t. 

Enough for dating, let’s move to the antivirus vertical. 

The advertiser here is split testing the angles “Your PC might be unprotected” vs “Activate McAfee Antivirus”. I personally like both, but they should use only one image for both creatives. 

We have a similar case here: 

Good, aggressive angles are ideal for the A/B test, but as we said, different images, icons and CTA buttons must be used. The problem is not that the images are bad, it is that, when a creative is performing better than the other one, we won’t be able to tell the reason why, because there will be too many differences between the two of them. So in order to find out why a creative is underperforming or overperforming, we need to test small changes at a time. 

I’m sure you already got it, and from now on you’ll do better split testing your creatives. 

Let’s now see the landing page! A lot of us already know there are only a few types of landing pages - a survey, a fake blog, a fake advertorial, etc. Yes, you can change the texts, images and product, but the idea behind all of them is still the same. 

Now let’s see some CBD offers landing pages in order to explain how to split test properly: 

I’m using a survey and an advertorial page - both are good pages and you can make a lot of money with them, Let’s take a look at them closely: 

There are different angles here: “do you suffer from chronic pain or stress” on survey nº1 and “retired doctor whose girlfriend experiences crippling anxiety discovers nature’s miracle and helped her within days” on the advertorial. 

If you want your A/B test to be properly done, you’ll need to choose only one and use it with the bot landing pages. Same about the images: the advisable variant is to use the same image on the creative too. Only then you will have significant data and a winning landing page. 

Pro-tip: When you split test your landing pages, also optimize the loading times for both of them, otherwise there is a chance the test will not be adequate!

And that’s how you can make an A/B test for all types of landing pages.

And, what about the offers? 

Why shouldn’t you test multiple offers in one single campaign? I’ve heard this question multiple times, and the answer is simple - because their performance is different. Even the same offer, but two different affiliate networks can show mixed results. It depends on their servers and algorithms, the landing pages they use and how many redirections they do, etc. 

Pro-tip: always ask your Affiliate manager which landing pages are the best for the offer and split test them too. The results can surprise you!

And how much time will this A/B test take?

That’s a very good question. Unfortunately, I can’t give you an exact answer! Why? Because your best bet is to continue with the split testing during the whole campaign - once you have found your best angles and images, create new creatives with them and test against the previous ones. Once you discover the landing page that brings you more money, change the CTA button or the color and again, do the A/B test Once you get the offers with the best EPC, try to find another one and start the test again. 

I understand this is time-consuming and only a few affiliates do this, so stop when you have a profitable campaign with 2-3 creatives and one landing page.

The most important thing that you need to remember about the A/B testing is: test only 1 or a maximum of 2 elements per creative and landing page, because if you test more components you will not know exactly what it is that’s working. 

Also, make your decisions only based on data, not because of feelings! 

Before saying goodbye, here’s a present for you - use this calculator if you’re not sure about your “winning element” - use it for everything - creatives, landing pages, or offers.  

I’m giving you a fast example:

A, B, C, D, E, F are the creatives we are testing, the “samples” are the clicks on them and the “Conversions” are the conversions they made. When pressing “calculate”, you got these results: 

In this case, I’ll stop the 2nd, 3rd and 4th creatives and continue with the 1st, 5th and 6th, because most probably they’ll continue to give me money. 

That’s all for today! 

Good luck with your A/B tests and do not forget to ask your account managers any questions related to the creatives.

Download PDF

Recent articles