When you build a website, you aim for perfection. However, what you consider the “perfect” website may not be in line with what your visitors will respond to — and might even drive them away.
Something as major as the hero image you choose for your landing page or as minor as the color of the call-to-action button can significantly affect your conversion rate.
That’s why it’s so important that you conduct A/B tests when you make your own website. Similar to the science experiments you did in high school, when you compared the effects of two different liquids on plants (was there ever really any doubt that water is better than soda?), A/B testing compares two different versions of your website to determine which one leads to more conversions.
By conducting A/B tests, you gain insight that allows you to make the right tweaks and changes without guessing or making assumptions about what will work.
A/B Testing Basics
A/B testing requires developers to work closely with marketing to determine what should be tested. Because it’s important to only test one thing at a time (testing too many different aspects of the page at once can skew the results) you need to work together to determine priorities. Some of the most common testing points include:
- Colors. Changing the color of the CTA button, headlines, text, and other aspects of the site can influence conversions.
- Headlines. Which version is more effective?
- Images. Do people respond to different images differently? Do they prefer images of people, landscapes, products, or conceptual images?
- Navigation. Do visitors respond to one word or phrase more than another? Does moving things around lead to more conversions?
- Links. Try using different language in links to see what gets the bigger response.
- Offers. Even if you are offering the same thing, try varying the language of the offer to see what gets the better response. If you have several offers, compare the results from two different, but equal, products.
- Amount of text. Do people respond better to short messages with more images, or does testing show more text is needed to get people to convert?
Again, choose one point to test at a time to maintain the integrity of the results. Otherwise, you won’t know if it’s the color or the image that lead to conversions.
Setting Up Your Tests
Once you’ve determined what you want to test, you have several options for setting up the test. You can either set it up so that the element being tested will be randomly replaced before the page is loaded, or have visitors redirected to either the control or test pages when they visit your site. Either of these options work, it just depends on your preference and the testing tool you’ve selected.
What’s more important is that you tie your test to a specific goal. To do this, you will need to insert some code into the testing tool that indicates a successful test. For example, if you are trying to increase downloads of an e-book, the goal would be the “thank you” or the download page, and the successful test would indicate that the user reached that point. By inserting this code, you can easily see which version (if either) was more successful.
It’s important to note that not every A/B test comes up with a clear winner. Sometimes, the results are all but equal or the results are not statistically significant, while other times there is no change at all. In order to ensure that you get results that you can use, you need to test the two versions simultaneously (don’t test the control, then try to test the second version) and that you give the test enough time.
There are several online calculators that you can use to determine how long to run the test, based on your current and desired conversion rate, average number of daily visitors, and how many visitors you want to include in the test. Depending on your goals, you may need to run the A/B test for a few months to achieve usable results.
Keep in mind as well that the results you get from your testing may not be exactly what you expect, or even think would work. You might hate the colors that users prefer, for instance, but remember: It’s not about you. The design may not be as aesthetically pleasing to your eye as you would like, but it could be exactly what your target audience wants.
If something seems really odd — a grammatically awkward headline, for instance — and you aren’t convinced that the test results are accurate, then consider re-testing, but be willing to go against your instincts if the numbers indicate you should.
A/B testing is an ongoing process, and you need to run tests continually from the moment you launch your site. Audience preferences change, trends change, and your business will change. By continually testing and improving, you’ll have a better chance of meeting your goals — and increasing your business’ profits.