There are many unknowns in online marketing.
Each offer is unique and each brand is different. Every target market is unique.
It doesn’t matter what works for someone else’s business, it might not work for you. The opposite may also be true.
This is where split testing comes into play.
Split testing can help you determine how to market, advertise, and sell online specifically for your business.
This guide will explain split testing and how it works. It will also highlight some of the problems that you need to be aware of.
Let’s get started.
What is Split Testing?
Split testing, also known as A/B or multivariate, is a technique used by online marketers to test the effectiveness of different elements of a marketing asset, such as sales copy, graphics, CTA, and CTA.
Split testing is when a marketer divides their web traffic equally between different marketing assets (websites, landing pages, advertisements, emails, and others).
… and then test to determine which variation drives the most conversions.
Marketers can either use the winning variant or continue to test and optimize.
Before we get into the details, here’s some split-testing jargon that you need to know:
Conversion – A conversion is when a goal is achieved (e.g. opt-in, purchase). The conversion rate refers to how many visitors ‘converted’.
Conversion Optimization – This marketing term refers to improving marketing assets to convert as efficiently as possible. Split testing is an important part of conversion optimization.
Variant – Split testing uses the ‘control’ version of the marketing asset. The ‘variant’ version is the one you believe will convert better.
Multivariate testing-Multivariate testing (or multivariable testing) refers to a more complicated version of split testing in which multiple variables are simultaneously tested to determine which combination performs best.
A/B Testing – Split testing can be described as A/B testing. It refers to the page variations of split tests, “A” and “B”.
You now know the basics of split testing and all the jargon.
Why split the test?
Split testing for online marketers: The benefits
Split testing has many benefits.
It allows online marketers to optimize their sales funnels to increase conversions, generate leads and make more sales.
Let’s take a look at split testing in action.
ClickFunnels has tested its “Top 1 Secrets Masterclass Webinar” to see if credibility badges would be more effective than a countdown timer.
The control version with the countdown clock won this case by a wide margin — 30% conversion rate to an 11% conversion.
ClickFunnels discovered that landing pages should have a countdown timer.
This lesson does not apply to all landing pages… but only to this experiment. We’ll get to that later.
This is the beauty of split testing.
Learn about the best and worst practices for a particular sales funnel or marketing campaign.
5 Marketing Assets That You Can Split Test (With Examples)
Let’s now discuss some marketing assets that might be worth split-testing.
1. Advertisements
Virtually all advertising platforms — Facebook, Google and Pinterest, LinkedIn, Twitter, LinkedIn, LinkedIn, etc. – are available. — allows marketers to split-test their ads.
The…
- Copy
- CTA
- Image
- Video
- Targeting
You can view all the Facebook page tests by visiting the “Page Transparency” section.
ClickFunnels’ ad library shows that there are 6 versions of the same advertisement.
You can look at the Facebook ads of your competition to get some inspiration.
2. Landing Pages
Landing Pages are ready for split testing.
You can try…
- Copy
- Layout
- Images
- Videos
- Countdown Timers
- Testimonials
- CTA
- Color
You can also find any other information on the landing page.
3. Checkout the Process
According to Baymard Institute research, cart abandonment rates hover just below 70%. This means that people abandon their online shopping carts without completing their purchases.
The checkout process is another asset that has the potential for split testing.
You can try…
- Copy
- Countdown Timers
- Buy Bumps
- CTA
- Flow
- Credibility badges
- Testimonials
- Loading bars
4. Upsells
Split testing is also possible for your upsells as well as your one-time offer pages.
Consider…
- The Offer
- The Headline
- The CTA
- The price framing
- Testimonials
- Loading bar
- Credibility badges
- Sales Copy
5. Emails
Email marketing is another area you can optimize and split-test. Consider…
- Subject line
- Preheader
- Sender Name
- Sender Profile Pic
- Email Copy
- CTA
Morning Brew, an email newsletter that has over one million subscribers, used split testing to make sure their subject lines got as many open as possible.
Here’s how Aweber explains it…
- Each morning, the Morning Brew team tests four subject lines. These subject lines are created from catchy and interesting phrases taken from the daily news.
- Each subject line is sent to 3% of their audience by 5 a.m. (12% total).
- One hour later, the winning subject line is sent to the remaining 88% of subscribers.
Split testing: The problems that no one talks about
Split testing can be a simple way to increase conversions.
It’s partially true.
Split testing has its dark side. There are three main problems.
1. Significance of Statistics
Imagine you are running split test and 200 visits randomly go to each landing page variation.
Version A receives 50 conversions and version B 40.
This is quite compelling, right?
It’s possible to be tempted even to ditch Version B and go for Version A. However, statistics show that you can only be 54.52% certain that the difference is real, which is less than a coin flip.
Also, your results are almost as likely to be random due to page variations as they are because of randomness.
Many online split testing claims have statistical significance that can be a problem.
Marketers often get too excited about the results and forget to consider the statistical significance. Then, the data gets shared even though it is flawed.
Many great online tools can help you assess the significance of split testing results.
2. Conversion Attribution
Marketers aren’t scientists.
Peer-reviewed scientific studies can be rigorous in eliminating confounding factors. Marketers sometimes make a variety of changes to the variation, then see an improvement and claim that one change was the difference.
However, it is impossible to determine which element caused the increase in conversion if multiple things were changed at once.
This is not a big deal for marketers. Do you care about how the conversion rate changed?
Most don’t.
It is important that you noticed an improvement.
However, you should be aware that split-testing claims made by other marketers might not be as straightforward as they appear.
3. Specific made general
Here’s an example of a well-known test that marketers have used online a lot: the red button beat the green button 21% to 1.
Is that important?
No.
It doesn’t. Because it may say something (assuming it’s real), but that landing page and that traffic… says absolutely nothing about you business, or your marketing resources.
This is a problem in split testing.
Split tests are a popular way for marketers to find out which element performs better than the other. It’s not that simple.
Some changes will work better on mobile than others on the desktop. It all comes down to the offer. The brand, traffic, advertisement, and many other factors all make a difference.
This means that what works for one funnel might not work for the next.
Top 3 Split Testing Software for Optimizing Online Assets
Split testing is not as simple as creating two versions of your landing pages, and hoping for good results.
Software that can randomly distribute traffic to your site, track conversion rates, and deliver results is essential when running split tests.
These are the top three to be aware of.
1. Google Optimize
Google Optimize allows you to run split testing on your website. You can use the features of Google Optimize to perform A/B testing and multivariate and split URL testing. There is also a visual editor, advanced reports, and integrations with other software.
2. VWO
VWO is an A/B test tool that thousands of brands use around the world, including Ubisoft and AMD. This tool is extensive and has many different features. This tool is for you if you have the time and desire to dive into split testing.
3. Crazy Egg
Crazy Egg software was created by Neil Patel. It allows users to create heatmaps and snapshots of website pages, as well as run split tests. This tool is extremely user-friendly and affordable.
9 Steps to Create a Split-Test That Is Useful
Whew!
Now you know the basics of split testing.
Let’s now dive into the 9-step process of creating a meaningful split test.
1. Take stock of Bandwidth
It sounds easy to create a high-quality split test.
Once you get started, it can be complicated. You will need to create new graphics, write new copy, build new elements, etc.
It takes time, effort, communication, and patience.
Split testing is a time-consuming activity. The more bandwidth you have, both for yourself and your team, the more split tests you can run and how much you can learn.
Split testing can be difficult if bandwidth is limited. Don’t stress about finding the cause of conversion improvement or reaching statistical significance at 95%+. Do the best with what you have.
Your sales funnel can always be optimized. This is where you set realistic expectations about what you and your team can accomplish.
2. Understanding Why Split Testing is a Good Idea
Why would you like to take a split test?
Split testing just for the sake of is not always a good idea, especially if you have limited bandwidth.
Make your split tests count.
Split testing is a great way to increase conversion, open rate, and click-through rates.
Ask yourself the following question: Is your sales funnel converting at an acceptable rate and profitable?
If so, you can put your split-testing budget elsewhere.
Split testing is a better way to determine which elements of a sales funnel are most effective. You can use heat maps and visitor recordings, as well as visitor recordings, to see how visitors interact with the product. This will reveal what they click, where they look, and when they drop off.
This will give you an idea of which elements must be optimized and tested.
3. Make a Hypothesis
Once you have a reason to test a particular part of your sales funnel it is time to develop a hypothesis.
You are trying to solve a problem. A guess as to what might be better than the current version.
It is possible to hypothesize that a countdown clock will increase conversion rates via FOMO and urgency. You might also think that different testimonials will prove more convincing social proof.
Before you conduct your test, make sure you understand your hypothesis.
If you are having trouble coming up with solutions that work, take a look at the strategies of your competition.
- What do they do differently to you?
- What is the difference?
- Why might they have a higher conversion rate?
Talk to your team, review the evidence from previous tests and CRO audits, and make an educated guess. Next, move on to the next!
4. Decide the size of your sample (and your budget)
You want to avoid tests that produce insignificant results as much as possible.
Jessie Valle (Director of Marketing at ClickFunnels) once talked to me about how fascinating a particular A/B testing they had run was. She responded,
It is fascinating, but even more interesting is the fact that more tests fail or are null. Unfortunately, there are not many winners.
That’s the essence of the game to a certain extent.
This is how to increase your chances of obtaining meaningful results. Before you start, determine the size of your sample and budget.
Use this calculator and work backward.
Enter numbers based on how you think the test should perform — and find out the statistical significance. You’re probably in good shape if you score over 90%.
To determine your budget, you can also estimate how much it will cost to drive the traffic required.
Although it is not a guaranteed solution, this will give you a greater chance of success.
5. You can create multiple versions
Now you’re confident that you have made the right decision and can determine the budget and sample size.
It’s now time to make your versions.
This is the most straightforward way to accomplish this: to only change one element on the page such as the headline or CTA or the opt-in form location.
You can test only one element at a time to connect changes in your conversion rate to a particular element.
This is the problem: if you don’t have a large budget to drive a lot of traffic it can take a very long time to make multiple optimizations.
After three months of testing, you will see a 5% increase in conversion. Then, test another thing for three months. And so on.
It’s tedious.
If you don’t have the bandwidth or budget to split test, you might consider multiple modifications to your variant in a single test.
This will force you to pay less attention to how impacts the conversion rate of your marketing assets and more to the impact. This is a way to consistently improve your marketing assets’ conversion rates without having to be so scientific.
Both approaches have their pros and cons.
6. Eliminate distracting variables
You can choose to change one element only or all of them. However, your split testing results will still be meaningful if you keep the same things consistent.
Specifically…
- Traffic Source
- Advertisement
- Load Speed
… and other factors not relevant to your hypothesis.
Two advertisements drive traffic to different landing pages. The problem is that it’s impossible to tell if the landing page or the advertisement has an impact on the conversion rate.
Although one ad-landing page pair might perform better than another, it is still possible to compare apples with oranges.
You don’t learn anything and can’t use the results for advertising decisions.
You should not use multivariate tests… but that’s another blog.
7. Test Your Test
Make sure that everything works before you start your test.
To ensure that you are using the correct software, visit the URL several times. Convert on both pages several times to ensure that the software is correctly tracking and reporting.
8. Drive traffic
Now, let’s inject some traffic into your test to make sure everything is in order. Take Step 4’s estimated budget.
Also, ensure that traffic comes from the same source with the same targeting.
9. Analyze the Results
What was your test score?
Did your hypothesis prove correct?
Are your results statistically significant?
These are the questions you should ask yourself as you examine the results. Congratulations if things turn out well and you can see a clear winner.
Now you can make the winner your control, and then let it be. Or keep testing and iterating.
Don’t be discouraged if things don’t go as planned. It happens more often than you might think.
Find out what went wrong, and then try again. Sometimes testing is necessary.
Conclusion
Split testing is valuable because it allows for the decentralization of optimization.
You don’t need to hope that the marketing strategy that worked for your competitors will work in your business.
You don’t need to rely on the wishes and whims of salespeople and marketers. It is possible to test your hypothesis.
Now you are at the beginning.
Have fun testing!
