A / B testing, also known as split testing, is a research tool that any self-respecting marketer should own. Using it, you can understand how effective your marketing efforts are, and improve them based on the findings. Surely, many of you know the method of split testing firsthand, and not once used it in their practice. However, the rest of this term still causes misunderstanding.
So what is A / B testing? This is a research method in which marketing variables are compared with each other in order to determine the option that finds the greatest response from the audience.
As the name implies, we are dealing with two variables: A (the original or the control version) and B (the new or the final version). They are alternately demonstrated to two homogeneous groups of potential customers for a certain time, until a high degree of statistical confidence is achieved. Then, the results are compared to determine the most advantageous option. In most cases, researchers assume that option B will change the conversion level for the better (although such expectations are not always justified).
The scope of A / B testing in Internet marketing is extremely extensive. You can test variants of landing page headers, lead-forms, images, banners, trade calls, etc. In our article we will dwell on three marketing channels (landing page, e-mailing and call to action, located on the selling pages of the site), and also give examples of the corresponding A / B tests, which were conducted by HubSpot marketing specialists.
The landing page is perhaps the most popular and favorite A / B testing site for internet marketers. This is due to the fact that it consists of many elements: a title, a subtitle, a text, a call-to-action button, an image, a lead form and some others. Moreover, each of the above components can be explored from different angles. Take at least a headline: you can experiment with its wording, size and font color, layout, etc.
As a rule, A / B testing of landings is carried out using special applications. Here are the most popular examples of such tools: Google Analytics, Visual Website Optimizer, Optimizely, Unbounce, etc.
Some marketers prefer not to split the landing page into separate parts, and see it as one variable. Therefore, when conducting an A / B test, they simultaneously change several parameters of their landing.
A similar test was conducted by the HubSpot team at the landing page optimization conference. The following initial page was offered to the participants of the experiment (a group of representatives of Central Asia), and they had to make changes to it, which, in their opinion, will have a positive effect on the level of website conversion:
But the final version, resulting in a series of transformations:
As you can see, the page has acquired a completely new look. First, the image has changed, as well as its position on the page. Secondly, the text has become much shorter. The length of the lead form has also been shortened, as well as its location. Third, the wording of the title has changed.
Given the scale of conversions, one could expect impressive results in higher conversions. However, in reality everything turned out differently. The results of the A / B test showed that the conversion rate increased to an insignificantly small value - only 0.33% (it was 47.91%, it was 48.24%). Why did this happen?
Lesson for marketers: optimize only those landing pages that have a really low conversion rate.
In this case, the A / B testing failed due to the following reason: the audience was already interested in the marketing proposal (to get free material), so it did not attach much importance to other elements of the landing page: text length and lead form, their location, etc. Hence the conclusion: conduct A / B tests only on landing pages with a relatively low level of conversion, it’s not so easy to compete with landings, which already convert more than 47% of the total traffic.
Call to action
The call to action also provides marketers with plenty of room for experimentation. You can change both its content (for example, invite your readers to subscribe to a new newsletter) and the form (use a different color and font size, change its position on the page).
For example, HubSpot specialists decided to change the validity period of their marketing proposal and see what came of it. Initially, they offered a free trial version of their program for 7 days. They wondered whether the number of their subscribers would increase if they extended the offer to 30 days:
The test results showed that extending the free offer to 30 days attracted a lot of new visitors. The level of site conversion increased by as much as 110% (from 32.6% to 70.9%). As you can see, the “generosity” of the HubSpot team paid off.
Lesson for marketers: start the call-to-action optimization process by testing your offer.
After analyzing this A / B test, we can formulate the following conclusion: the type of your marketing proposal can have a tremendous impact on the lead generation process. So start optimizing your calls to action by matching different marketing offers. Such experiments will help you to better understand what exactly helps to turn visitors to your site into leads.
When conducting A / B tests, e-mail marketers most often compare different variants of the subject line. And this is not accidental, because a correctly selected subject of the letter can attract the attention of recipients and have a positive effect on the level of clickability. However, using split-testing can explore other aspects of emails, such as the sender's name.
In 2011, the company HubSpot decided to check whether such a seemingly unimportant nuance, like the name of the sender, can increase the effectiveness of e-mail distribution. The name of the sender first indicated the name of the company “HubSpot”, and then the specific name of the specialist from the marketing department. That's what happened in the end:
The graph shows that the level of clickability on the link in the original version of the message reached 0.73%, and in the final version - 0.96%. So we have the undisputed winner - option B (statistical confidence is 99.9%).
Lesson for marketers: your email is more likely to attract customer interest if you send it on behalf of a real person, not a company.
Perhaps here and everything is clear. People trust more letters received from a particular person, and not from an abstract company. They want to know who exactly is entering into a dialogue with them. So why not satisfy their desire?
Test, draw conclusions and ... test again
A / B testing is a multifaceted and rather unpredictable research method. It happens, you think: “This title will definitely“ blow up ”the conversion level on my site!”, But in reality everything turns out differently. At best, you get modest results, and at worst - no. So you should not rely on your subjective opinion - it is better to check everything with the help of A / B testing, which will show you the true state of affairs.
The most important thing - do not stop there. Suppose you moved the call-to-action button to the “first screen” zone, due to which your site’s conversion rate increased by 50%. This is just great, congratulations! Now, you can experiment with its color, for example, change it from gray to bright yellow. And you do not have to be limited to the site: A / B tests can be carried out on pages on social networks, and in a blog, etc. As they say, there is no limit to perfection.