How to Set Up Your First Website Split Test
Split testing is the common thread in the world of digital marketing.
We all know we should do this, but we don’t.
Worse, many people who share the test could be wrong and do more harm than good.
I’m not going to talk about the importance of split testing because it’s fair. Instead, I want to help you start your first test. If you’re new to the game and an experienced vet, I want to show you how to properly set up your tests.
page selection page
This is probably the most important step in the test creation process. If you choose a page that should not be tested, you will have a bad test.
Tests should be selected based on probabilities.
Remember that happiness doesn’t mean fixing your worst pages. This means testing the pages that have the greatest impact on completion.
In other words, the page you choose can be a big boost and should point directly to your key progress indicators. Here are some examples of how to pick the wrong page:
I worked for a company that wanted to test an email campaign for email campaigns. I asked for the page conversion and they told me it is 75%. Yes, 75%! It was a large B2B company looking for more qualified marketing contacts (MQL).
My answer was “The conversion rate can’t be 75%”, but let’s assume it’s correct. The team is clearly asking the wrong question. The landing page should not be tested, but the email campaign that drives traffic to the page should be tested.
To identify your tests, open Google Analytics (or your preferred analytics platform) and start searching for data.
One of my favorite brands is payment pages. If you have a page with a different high drag speed than the bottom of the page, you have identified a possible test page.
Which of the four identified output pages should not be a test priority?
The completed payment page does not need to be tested!
This is the last page of the email exchange and should not be a priority. Sure, you can try adding more upsells to this page, but it’s not as important as these other pages!
Of these four pages, I recommend viewing the shopping cart. The conversion value is easy to quantify and the traffic is highly qualified.
Now that you’ve done all the work to identify your Page, get ready to…
A little more work.
This is the funniest part for our less tech-savvy readers!
Now you need to identify the evidence or what you expect to increase conversions on your page. Please don’t guess. The random selection of elements transforms the testing process into a game of hypothesis and verification.
Check out some quality data sources like eye tracking, click tracking, user surveys, etc.
Here is an example of a heat map showing the average number of user clicks. Note that people click where they want. When checking click tracking, make sure your most important call-to-action isn’t in the shadows!
This will give you an idea of which aspects of the page are not working for your customer. Remember that testing is meant to optimize the customer experience; a test should not be based on what you think works best: you are not your customer!
If you don’t have access to good-quality data, you can refer to test cases from companies in your niche. If you are a super niche and the only one who ‘does what you do’, try ‘best practices’.
Once you have your page and its elements, put your idea under a test hypothesis.
Okay, that probably makes people nervous.
These technologies include Optimizely, VWO, Marketizator, and Convert.com.
I know it looks good, but copy the code and send it to your developer or paste it between the <head> … </head> tag.
Once the code is added, this is it. Simple, right?
Step by step example
Let’s put it into action.
In this example I use VWO, it is not a product approval. Find out which tool best suits your needs. I recommend that you read the TrustRadius Technical Test Report before making any technical purchase.
Step 1 – Choose the campaign type
Most testing technologies cover different field types, including A/B testing, URL breakdown testing, multivariate testing, and other qualitative testing campaigns. In this example, I select the A/B test.
Step 2: Enter the test page URL
Before opening the test technology, you must have identified your test page. When you are ready to use it, enter it in the URL field.
Step 3 – Start making changes
Test tools are provided by a WYSIWYG editor. For example for small changes. Title, call to action, images, etc… the tool should work perfectly. Other large-scale operations, such as large-scale drawings or modeling changes, are not suitable for this editor.
Step 4: Determine the goals of your goals
During the testing process, I asked you to select the test pages that match your KPIs. You can choose different targets and use them in different ways to control them. The most common is to enter a destination URL, probably your thank you page.
Here is a list of the different ways to track targets with VWO:
Step 5: Complete the test
All you need to do is name your campaign and indicate the amount of traffic you want to send for this test.
I always recommend scheduling a test for a period of time. Don’t do an ad hoc test; you can calculate its duration using a calculator. Once you have the test date, you must complete it to complete the week and it will be completed on time!