Kajsa Hedqvist & Elin Ledenvik
24 September 2020
Do you have trouble getting your A/B tests prioritized by the development team? You are not alone! Find out if Redirect tests, or redirection tests, can be a solution for you.
One of the biggest bottlenecks when it comes to A/B testing is development. It can be difficult to get tests prioritized in the development queue, there may be a general lack of development resources, or it may just be a temporary freeze on development for various reasons.
However, there are some ways to A/B test without needing development help. Many A/B testing tools have a visual (WYSIWYG) editor where you can make simple changes, which you may already know, but what we will look at in this post are so-called Redirect tests, or redirection tests.
Being able to test without the involvement of the development team is not an excuse to just run simple, spontaneous tests. It is still important to develop hypotheses based on data and insights and to find a way to prioritize your ideas to test what has the greatest potential.
The redirection method, however, gives you the opportunity to test content such as and other content that can actually have a significant impact. Additionally, the best way to get an A/B testing process started might be to just get a test out there to show how it works and what results it yields.
Examples of hypotheses we have tested with this method:
đź’ˇ Address visitors' needs earlier on the page with an added Call-to-Action button in the hero image.
đź’ˇ Facilitate product comparison by displaying product cards in a vertical list instead of having to scroll sideways on a listing page on mobile devices.
đź’ˇ More clearly answer visitors' questions on a product page through a redesign with a new information structure.
Note: Even though this method allows you to test without relying on the development team, it is important to know why you can't get tests developed in the first place. For example, is it due to a temporary freeze on development because of vacations where no one will be able to stop the test if something goes wrong, or a pause due to a major release or campaign? So make sure to consider this even for this type of test, and only start the test if you know there are conditions for a successful result.
1. Sit down with your Content Manager(s) to get an idea of what changes are possible to make in the system you use to manage your web content, your CMS (Content Management System). This could be, for example, WordPress, Episerver, Adobe, Magnolia, Contentful, etc.
This can be done either before you create your hypotheses, to know what limitations exist in advance, or after hypothesis creation to see how you can test a certain hypothesis by only changing content via the CMS.
2. Make a copy of the page you want to test in your CMS, with a slight change in the URL. For example:
https://conversionista.com/tjanster/ab-testning/ → https://conversionista.com/tjanster/ab-testning-1/
3. To avoid harming SEO, the new page needs to have either a canonical tag or a noindex tag. These are used to ensure the new page is not indexed by search engines, as this could negatively impact your ranking and lead people directly to variation B. Both tags are good, but it is usually not necessary to use both. Sync with an SEO colleague.
4. Make the changes you want to test on the new copy of the page. Make sure it looks and functions as it should to test your hypothesis.
5. Do a QA (Quality Assurance) by visiting the new page from different devices and browsers. Test clicking on links on the page to see that everything works as it should. Since you haven't made any code changes, there shouldn't be any issues, but you should always do a good QA.
6. In your A/B testing tool, set up a test to redirect 50% of visitors to the new page. It is often called Redirect test or Split test in the A/B testing tool.
7. Once the experiment is created, click on Add variant. Then select “Redirect to a single page” and add the URL of the new page that will be the VARIANT.
8. Set up the rest of the test as usual, with the goals (objectives) and any audience targeting criteria that you want to have.
Page targeting should be set to the original page.
9. When you are done with the setup, click on Preview to test that the redirect works and that you can see the changes. If you can't see any changes, choose Debug instead in the preview menu to see if the targeting rules are correct. If the experiment works as it should, you will only see green icons; otherwise, you will see an error message saying “This experience was not applied” and what is wrong. Fix the problem and try again.
10. Start and run your test as usual. When visitors come to your original page, the test tool will automatically send 50% of the traffic to the variant's URL instead. This way, after the test, you will be able to see if your new page with the changes performed better than the original or not.
"When visitors come to your original page, the test tool will automatically send 50% of the traffic to the variant's URL instead."
If the test is a winner, it will be easy to implement the changes by simply making the changes on the original page in your CMS.
Remember that the only bad test is the one you don't learn anything from. Did the original perform better? Then you have tested something that you see affects user behavior and can start digging into why. This way, you create new hypotheses based on your tests.
"Remember that the only bad test is the one you don't learn anything from"
By changing content in your CMS, you have the opportunity to do A/B tests without needing help from developers. This way, you can get A/B tests out without having to wait for the tests to be prioritized. However, relying solely on this method is not sustainable in the long run as your testing will be limited by the changes you can make yourself and not by the changes you think can have the greatest impact. The goal is also to get the entire organization on board with the CRO train.
However, there are many changes you can make with this method that can have a significant impact, as long as you ensure you have a well-founded hypothesis.