The Ultimate Beginners Guide to Affiliate Marketing Split Testing

All affiliate marketers (without exception) have one shared goal: make more money.

And to do that, everyone has to:

  • Increase traffic from a variety of sources (organic, paid, etc.)
  • Improve conversion rates for existing traffic

You can find some great ideas on how to increase your traffic from our new Affiliate Spotlight section, where affiliates just like yourself share their trade secrets!

As for improving your conversion rate, read on to see how you can double your earnings with one of the oldest tricks in the book – Split Testing.

The basics: What is Split Testing?

Split Testing (or A/B Testing) is a technique used to compare the performance of two or more versions of a web page (if more than two, you call it a Multivariate Test).


Split testing webpage


This “performance” is calculated based on a particular page goal or objective:

  • Product reviews page goal could be the usage of an affiliate link
  • Lead capture page goal is to get leads (subscribers)
  • Contact us page goal is the submission of the contact form
  • Homepage main goal might be to get clicks through a banner or call-to-action
  • And so on!

Knowing these goals, you are able to set up tests on a page (original version vs new variation) in order to improve it’s performance (conversion rate) and increase your earnings.

Note: Don’t get confused here! In affiliate networks you will see conversion rate reported as sales per click that you send to a particular merchant, but conversion rate is a generic term that can be applied to virtually any goal (sales, clicks, subscriptions, etc.)

Let’s get started: what should you be testing?

The answer is simple: what your visitors tell you to change!

At the end of the day, it doesn’t matter whether you like your website or not, because you are not the one who is going to be using it.

Obviously, you do need some kind of common sense when building and improving it, but it doesn’t need to be the top model of all affiliate websites and it SURELY doesn’t need to be a page full of 90’s gif files simply because you love flashing disco lights.

Here’s my three “must-tick” boxes for any website:

  • Informative content (the core of any good website)
  • Easy to navigate (good navigation and site architecture)
  • Fast loading speeds (remember, mobiles are ruling)

Looking at these, you should be asking yourself the following:

  • Is my content informative? Do I miss any information on product reviews?
  • Is my site easy to navigate? Can visitors find what they need in a timely fashion?
  • Is my site loading fast or does it take more than a few seconds?

When it comes to the two other questions, there are multiple ways to get REAL answers for them, coming from actual visitors or other people who are in the same knowledge level.


Google Analytics (Real Visitor Data)

Google Analytics (or an equivalent tool) is the single most important source of data for every website, because it gives you SO much information. And the best part – it does it for FREE.

Using Google Analytics you can get a massive amount of data from your visitors without asking them. It’s also very easy to read and understand for anyone.

Note: If you aren’t doing it yet, get yourself a way to track affiliate links clicks as goals. For WordPress you can track outbound links with a plugin called Monster Insights Analytics (previously known as Yoast Analytics). And to get even more data,  I suggest you look in your MoreNiche affiliate area for your overall EPC in the last 30-60 days and use that as the revenue value of each goal (click). If you have doubts about this integration, feel free to comment below and I will help as best as I can!

Anyhow, I am not going to delve into all the cool things you can see in Google Analytics (that’s for your free time, to learn and explore), so here’s the report I want you to look at right now:


Behaviour → Site Content → Landing Pages

Google Analytics Data


This is the single most important report you have. It allows you to see how each of your pages perform and get usable insights from that. Some of the stuff you can learn from it include:

  • Bounce Rate: This shows the percentage of customers who only visit a single page on your website. Your product reviews will probably have an high bounce rate and that’s not a problem as people might be leaving through affiliate links. But generic pages or pages where you expect customers to land and then interact / navigate elsewhere with the site should have a lower percentage.
  • Average Page Per Visitor: This is exactly what the name says. You can analyse both this and your bounce rate together, as they are somewhat connected.
  • Average Session Duration: Unless your site is on the very last step of a customer journey, having longer session durations is an extremely good sign. If your customers are leaving after a few seconds, perhaps you can test a variation with a video or more engageable information at the top to grab their interest.
  • Affiliate Link Conversion: If you don’t have this, see the note above. This value will show when you have the goal correctly setup for outbound affiliate link clicks. With it you will be able to see money making pages and traffic wasting ones.


Using the affiliate link conversion goal you should be comparing similar landing pages as much as possible.

Here’s an example:

Your PhenQ review has 30% CTR on affiliate links, but your Proactol XS review only has 15%.

Based on this, you should be testing a different variation of the Proactol XS page with different text or image CTAs (probably similar to the successful PhenQ ones).

Ultimately your goal is to increase your affiliate clicks (same traffic, more money!).

On-site Surveys (Real Visitor Data)

You have probably came across on-site surveys before. They are a simple and effective way to get real visitor data. We have recently used this one on Brutal Force product pages:

Web Question


You can ask different questions (just ensure they are short and snappy) but the end goal is the same: to get real visitor information that you can use.

Using the question in the screenshot above (set it to show on exit intent) you would be able to understand what information might be missing on your product reviews, such as:

  • More information on ingredients
  • Shipping locations
  • Discount savings
  • Coupon codes available
  • What products to combine/stack with
  • If X product is suitable for males/females
  • And so on

The answers you will get are limitless and while you might get some random answers, the big majority will be extremely useful data that can have a massive impact on your revenue.

Imagine that you saw a pattern (e.g: customers asking for what products to combine PhenQ with). It could well be that customers are leaving your website and looking for this information elsewhere (other affiliate sites etc) and that means you are losing out on a sale.

In this case, it could mean that simply adding it would fix your problem, BUT, what if adding this means you are giving your customers more distractions (e.g: researching for the product you suggested combining with)?

Here’s where the split testing will come into place. You should test a new page copy with the extra information and see how the performance changes.

Since you are basing this on real data, your hypothesis is very strong and as such it is more likely that this will result in a win.

Recommended Tool: You can run on-site surveys for free with a tool named Hotjar. Besides on-site surveys you can also use it to set up heatmaps and user recordings to get even more insights and testing ideas.  

Usability Tests (Approximate Visitor Data)

As a third data source, you can run Usability Tests.

These are basically virtually set up tests where you ask your user a bunch of questions while they are navigating through your site. An example of some tasks/questions would be:

  • Put yourself in the mind-set of someone wanting a diet pill to lose weight
  • Open mysite.com and look for the best diet pills in the market. Were they easy to find?
  • From my top list, which products stood out to you and why?
  • Move on to the review of your chosen product. Is it what you were expecting to see?
  • What do you think of the information provided? Is there anything missing?
  • If this wasn’t a test, would you be ready to buy X product?


Again, this is just a quick example and you can find more simply by googling “usability testing questions”, but hopefully it helped you understand the basic concept.

The best part of these tests is that you can actually see the user moving through your website while they are answering your questions. You will likely have moments where you will be yelling at the screen and asking “why the heck has he done that?”.

Now, there are 2 kinds of usability testing: moderated and unmoderated.

The most common one is unmoderated.

You will simply use a tool (such: UserTesting.com) where you will setup a test for as many people as you wish, with a set of demographic filters. You will often get some test results in as soon as 24 hours (maybe more if your filters are strict). There is a cost attached to this (per video).

Not as common and just about as efficient (but more limited) are moderated.

For these you can come up with a bunch of test questions, print them up in a document and get a real person to sit next to you or in real-time on Skype.

You will then ask that person to open the website and run them through your questions, taking notes of their answers and of what they are doing on your website. If possible, you should also run a screen recording so you can review it in detail later on.

In terms of costs, you can pay for people to do moderated usability tests (finding them on sites dedicated to usability testing) on Skype or other chat tools, but you can also do it for free by asking your family and friends to sit down for a few minutes.

If you go with the latter (which is more than fine, and free!) make sure that the people you ask for these tests are not familiar with your websites and industry. The less they know, the more likely they will behave like a normal customer.

Which idea should you use for testing first?

The above methods will hopefully get you a bunch of nice ideas for testing, but now you have to do something else – prioritize them!



And to do that, there’s nothing better than good ol’ Excel (or Google Sheets). Here’s how you do it:

  • Create a spreadsheet with 5 columns:
    • Idea Description
    • Perceivable revenue improvement
    • How easy it is to implement
    • How easy it is to maintain after it’s live
    • Final value
  • Add the collected ideas to the 1st column
  • Score the ideas on columns 2, 3, 4 from 1 to 10 (10 being the the easiest and the ones that you think will give you more revenue)
  • In the Final Value column, multiply the values from the 3 previous (e.g: 8 x 6 x 10)

Once you do this, you will likely come up with numbers like from 100 – 1000 in the final column, which will be your priority list. Then, simply sort the list from largest to smallest and you will get a nicely prioritized ideas list for testing.

So let’s get on with it!

With the ideas prioritized, all that’s left for you is to crack on with your Split-Testing.

There are several paid tools available, but here’s a few that won’t require you to rob a bank:

Visual Website Optimizer (VWO) – Plans start at $49/month. We use this at MoreNiche. It’s a great tool with several good features and the support is top notch.

A/B Tasty Plans start at $34/month.

Convert – Plans start at $69/month.

Omniconvert – There’s a free plan for up to 10k visitors. I’ve never used it but might be worth a shot to see if the free features are enough for your needs!

And of course, last but not least, Google Experiments is free, but only allows you to test organic traffic. You will find this tool sitting right in your Google Analytics dashboard, under Behaviour -> Experiments.


Google Analytics Report

How do you setup pages for testing?

Some of the tools above have visual editors which allow you to create page variations by simply editing the copy and some layout bits in the tool and without you needing to change content on your website.

However, if you want to use Google Experiments and run split-tests with it you will need a new page for the variation.

Here’s a quick guide on how to do this in WordPress:

  • Create a new page / post, set it to no-index (to avoid duplicates) and hide it from your category pages.
  • Go to your original page/post (e.g: mysite.com/phenq-review) and copy its content.
  • Paste it in your newly created page/post
  • Edit the permalink of the new page/post to mysite.com/phenq-review-v2 or similar
  • Go into Google Experiments and create a new experiment
  • Set the page to test (original) and choose 100% of the traffic (faster results)
  • Setup the goals (affiliate link clicks, more time on page, etc.)
  • Select the page of the variation
  • Start the test and wait for it to have usable data


You will likely need at least 2-3 weeks of data to get somewhat accurate results for the micro conversions you are testing. Make sure you let it run for at least that amount of time (even more if necessary) to ensure you get a real result (positive or negative).


Got accurate results? Brilliant, let’s use them!

You probably heard that a test result is always good, regardless of whether it’s positive or negative and there’s some truth in that. Though, as everyone else, I am sure you are more keen on positive ones, since these are the reason why you are testing in the first place!

So, if your split-test was positive, the outcome is straightforward. Pause/disable the test and edit the original with the changes in the variation so you can enjoy the results of your learnings.

However, if your split-test was negative you should spend some time thinking about why it didn’t have the expected impact, if there’s something you should re-test or if you should simply scrap the idea all together. You will often find some things that might make you want to re-test and that can turn a negative test into a massive win the second time around!

That’s it, all up to you now!

I hope this little affiliate marketing split-testing guide is helpful and has given you a push when it comes to collect and test ideas to improve your site performance and increase your affiliate revenue.

Don’t forget to comment below if you have any questions or would like to share your crazy testing ideas!


Leave a Reply

Your email address will not be published. Required fields are marked *