When it comes to making landing pages, email copy or even ads, it can be hard to gauge what will perform the best.
Relying on guesses or chance for your marketing decisions is risky. You’re much better off split testing.
Most people have heard of the term A/B testing before. SEO split testing is relatively new to the marketing world but is quickly becoming an essential tool for conversion-driven companies.
What is Split Testing?
Our goal with SEO split testing is to compare multiple versions of the same ad or landing page and see which performs better.
Imagine you’re at the optometrist, getting an eye exam. Split testing is similar: you try out different variations until you’ve found the option that suits you best. When it comes to SEO, the best option is the one that most effectively achieves your goals for the campaign.
In order to begin split testing, you’ll need some ads. The best practice is to create two or more ads that have minor changes to the design, copy or layout (more on that later.) You’ll then show each ad to similar target audiences and monitor their performance.
The Benefit of Split Testing
Simply put, split testing is valuable to businesses because it’s low cost with a high reward.
You could pay someone to write five articles per week, but they may only generate 10 leads. Imagine the savings you could earn by only writing one article in the time it takes to write two but split testing the calls to action (CTAs).
You may find that the number of leads goes from 10 to 20. The extra time spent writing the article means it’s of high quality and unrushed. Even if the test doesn’t yield the results, you can use the knowledge you gained to make data-driven changes for the next time.
The eventual success of your tests will ultimately set you off better than a business that didn’t test.
It’s easy to determine which title has a bigger impact or which button is most clickable using testing. Minor changes and adjustments can make your conversions take off and keep people on your page for as long as possible.
How to Run Split Testing
Before You Run the Test
Step 1: What do you test?
Before you do anything, you have to identify the aspects of the ad or landing page you want to test. It may prove to be a longer list than you bargained for.
However, it’s a good idea to only one variable at a time. That way, you’ll know whether (and exactly how) the change made an impact. These changes don’t have to be major. Even changing the colour of your CTA can improve your results!
If you want to test multiple variables, that’s perfectly fine! The best practice is to test them individually and identify the top performers.
Now, there are times when it does make sense to test multiple variables instead of just one. This is called multivariate testing. This method takes longer to set up and requires more traffic to complete. The more you change, the more combinations you have to test to get useful results.
Step 2: What are your goals?
It’s important to identify the specific goal you want to focus on throughout the test.
Though you’ll end up monitoring multiple metrics for each test, you should choose a primary focus before you start testing.
Why? Simply because it allows you to identify which variable will influence what metric.
You should identify where do you want the variable to end up when the test is done.
Step 3: Creation
It’s time to create your various tests!
You have your variables and goals, so all that’s left is to make them. The first step is to make a control – an unaltered version of whatever it is you want to test. If this is a landing page, you would design and create copy how you normally would.
Once that is finished, it’s time to build your variant page or ad. This will be whatever you’ll be testing against your original. For example, if you’re wondering whether videos or images provide higher engagement, you would set up your unaltered version with just images. Your variation would then replace the images with videos.
Step 4: How Significant Do the Results have to be?
It’s easy to say that the results just have to be better than the original.
But by how much? If you got one more conversion than the original, is that worth it? Probably not.
Statistical significance is the most important part of the testing process. You may recall the term confidence level from your old statistics class. Typically you want to have a 95% confidence level, but the higher the percentage, the surer you are that you’ll have the results you think.
During the Test
Please note that you shouldn’t run more than one test for a single campaign. If you do, it can complicate your results. If you have an ad campaign that directs to a landing page that you’re testing, how do you know which one is actually generating leads?
Step 5: Test the Control and Variant at the Same Time
To begin your test on a website or email, you’ll need to use a testing tool.
Using Google Analytics’ Experiments, you can test up to 10 versions of a web page to monitor the performance.
Once you have that in place, you’ll need to run the campaign at the same time. If you were to run test A about furnaces in the winter and then test B in the summer, you won’t know if it was actually affected by your changes or if it was just the time you ran them.
Step 6: How Long Should the Test Last?
Oftentimes one of the downfalls of testing is limiting the amount of time to see the results.
You won’t see results overnight. It typically takes a week or two to see the variations in the different tests.
So don’t panic if you don’t see results yet!
After the Test
Step 7: Was the Test Significant?
It’s time to reflect on the test you just ran.
Looking back on your original goal, did you meet it? Which performed better?
Once you determine that, it’s time to find out if the test results were statistically significant. Can you justify the change?
Hubspot offers a free split testing calculator for you to use if you don’t want to do it manually.
Step 8: Improve!
These tests are all about making improvements to gain traffic and conversions.
So use the information you’ve gathered to improve your results.
If the test you just ran didn’t work out how you thought, then run another one with a different change! If you were successful, use that in your next campaign.
Use your tests to discover new ways to develop your own content and improve!