People are unpredictable, especially when it comes to getting them to click and buy your product.
Just because you think you have sent out the best ad you ever created, or you have written your most convincing copy, does not mean you will make as many conversions as you intended to.
What you think or feel will work may end not working. This is why it is necessary to run tests – more specifically, A/B tests.
What is A/B testing?
This type of testing is also called split testing.
It is a marketing experiment that compares two versions of a piece of content and tests them to see which one converts most.
You show a part of your audience version A and you show the other half, version B.
Let me paint you a picture:
Imagine creating an email campaign for your existing email list.
Let’s assume the email is about a promo, you can decide to create two versions of the email (changing texts, pictures, and even videos), then you send a version to the first half of your audience, and send the second version to the second version to the second half.
Watch closely how both emails perform for the next couple of days, and pick the one that made people follow through with your call to action.
You can test your web pages, blog posts, emails, ads, social media posts, – anything.
There are several elements on a page that you can split test. They include:
More importantly, if you want increased sales, you should test your sales copy and product description pages. Test the elements we listed above in these pages.
You can also test the design and layout of each page.
Why do you need A/B testing?
The simple truth is every business needs to split test their content. Here are some reasons why you should:
- Higher conversion rate
- Increased traffic on your website.
- Reduced bounce rate on your website
What benefit you get is determined by what you are testing and the goals you aim to achieve. But ultimately, split-testing gives you increased conversion.
How to run A/B testing
1. Test one variable at a time
The rule of testing is to test one thing at a time.
You can decide you want to run a split test on your whole funnel but you have to make sure that you test each stage of your funnel one at a time.
Also, if you are testing a page, make sure you take one element at a time. if you want a more detailed analysis, you test one variable of each element.
For example, you decide to test your sales page. You should not split test text, images, headline, CTA button at the same time.
So, let’s assume you pick the CTA button. Do not change the copy, colour or size at the same time – that is, if you want very detailed results.
Tackle the colour first, then the size, then the copy. If you do all three together, when it starts yielding results, you won’t be able to pinpoint which variable of the CTA button is causing the increase in performance.
2. Identify the goal you intend to achieve
Ask yourself, what do I intend to achieve from this page I want to split test?
When tracking performance for each variation, you will measure a number of metrics, but it is important that you choose one main metric to focus on.
You should do this before you even create a second variation of the element you intend to test.
This helps you run a split test that effectively triggers the right user behavior you want from your users or prospects.
3. Create a variation
In A/B testing, there are usually two main elements, the “control” and the “variation”.
The control is the first piece of content or element that you design while the variation, is the variant of the same piece of content you design in order to compare performances.
For example, if you are testing your landing page, your control is the one you are using already while the variation is the second one you create to challenge the first.
If you have text-based testimonials in the control, you can remove the text-based testimonials in your variation and include video testimonials instead.
4. Choose random sample groups for your split test
The next is to split your audience samples.
If you are creating a test where you have control over the viewers, you should test with two audiences or even more.
Split your audiences randomly and equally.
But, if you are creating a test for something that doesn’t have a definite number of viewers, for example, If you are testing a web page, you need to test long enough for you to get a substantial number of views.
5. Test both versions simultaneously
You have to test run both the control and the variation at the same time.
Timing is very important in getting results. The time of the day, day of the week can change results drastically.
So, make sure you run them at the same time. Put them out in the same environment and see which one flourishes the most.
The only time you can test differently is if you are testing for timing. For example, if you want to know the best time or the best day to send emails.
6. Run the test for at least a week
This is very important.
Running your split test for a week makes it possible for you to track each version’s performance, every day of the week.
You can even run the test for a longer period of time.
The way people interact with emails, social media, web pages differs with each day of the week.
For example, the type or number of people that interact with your site on a Wednesday will differ from those who do on a Sunday.
Running it from Monday to Wednesday will not include performances from the other days which is necessary for your overall conclusion.
But more importantly, keep running the test for as long as it is necessary to get substantial data.
7. Take action
All these are not important if you do not take the necessary action.
If at the end of the test, you realized one variable performed better than the other one, then you implement it as the main content.
If none of them outshines the other, you can mark the test as inconclusive and run another one. Keep the Control and create another variation.
The data from the failed test should help a creating your new one.
If you want increased conversions, one of the ways to get it is A/B testing. Just make sure you keep testing until you find out which piece of content performs better than the rest.
If you have any question or a tip to share, please leave a comment. I would love to hear from you.