What is A/B Testing, and how to conduct it. “Do not take anything for granted. Only decisions from practical results, that is – test results. Testing should be all and all. I do this all the time. ” (c) Gary Helbert.
WHAT IS A/B TESTING AND HOW TO HOLD IT
And in this with the King of selling texts (in other words, the most famous copywriter of our time), I completely agree. A / B testing is what you need to make the right decision.
The decision is based not only on their own feelings. What is it and how to spend it? We will talk about this in this article.
What is A / B testing?
A / B testing (also known as split testing or group testing) is a method of comparing two versions of a web page or application with each other to determine which one works best.
A / B test (or split test) is one of the marketing methods when one group of elements is compared with another group of elements with different data. The goal of the action is to find out from which group of elements the conversion or indicators will be higher.
That is, everything is quite simple. You create 2 different options for promotional products or emails selling your services. And at the end of these letters you make a different call to action. And just consider the conversion of a letter was higher. That is the essence of the a / b test.
Here is a vivid example at the site level, where we determine how the changed arrangement of blocks affects the result. Namely raised to the top of the range.
But if you are testing, for example, 2 landing pages, in which the headers, call-to-action buttons of different shapes and different sizes are changed, then this is not just a / b test, but multivariate testing or A / B / N-test.
So why so much is said about this test now? Everything is very simple. Marketers love to show their expertise through various kinds of chips, to which testing can be attributed.
And company owners dream of finding a magic pill and believe that split testing is simply a panacea for their promotional materials (in particular when creating the landing page).
WHAT WILL YOU GIVE THIS TEST?
In fact, not everything is so bad and the split testing of the site really provides benefits for companies. And it can change the situation in a short period of time, namely:
- Increase conversion. The easiest and most desirable indicator of any owner, especially in websites;
- Changes in behavioral factors. Not the most obvious, but again the factor affecting the increase in sales;
- Increase average check. This is the addition of a different kind of call to action or by the example of sites, the familiar “buy with this product” block.
To summarize, it all comes down to increased sales. And definitely worth your attention. But there is one “BUT”. And it is already possible at this stage you will realize that you do not need it.
This “BUT” says that to achieve explosive sales by replacing one element, if it is possible, it is very rare. Approximate chances are 1 to 1000. Since there is a replacement of small components that rarely differ drastically from each other.
Therefore, we recommend doing this kind of experiment after that when you increase the conversion to 1% will make the weather. And when for you this is a drop in the ocean, it is better to focus your attention on more important components.
In addition, as long as you do not have well-established traffic. While you are just testing different channels and ways of advertising on the Internet. You may not even start testing, since the indicators will not be correct. After all, traffic is not homogeneous, but, as you know, different people act differently.
TIPS FOR CORRECT ACTIONS
If you have no experience in conducting a / b tests. I highly recommend you to pre-read cases. Especially people who have achieved increased conversion / behavior change with a / b tests. So you will understand what is best to test.
Or you can go the other way. Make a preliminary list (plan) of what you will test on the site. Based on their assumptions, feedback is the focus of the group (the worst option). It is necessary that you do not throw from side to side.
In our experience, the bundle works best – 1 test = 1 change.
Important! Forget about multivariance and do only one test at a time.
Otherwise, you will not understand what actually yielded results. And it will be even more interesting when one change gives +0.5, and another change on the same page is -0.5 for conversion. As a result, 0, and the overlooked positively influencing element.
Important! Forget vague theories. From the series “it is necessary to change something in the website, making it more selling.”
The normal theory for testing – on one page the button is red, on the other blue. On one page there is a price, on the other there are no prices. Only then will you be confident in the results. Clear criteria = clear results.
Important! The result obtained in a very short time = bad result.
Especially if in a short time on your site has been on the strength of 20 people. Therefore, before you rush to the testing site, you need to understand how many people visit it daily and how many days the testing should last.
Test Time Calculator
So that you are not bored, I found a calculator, thanks to which you can calculate the optimal duration for testing the site –
Here you will need to fill in:
- Conversion of your page at the moment;
- The percentage by which you want to increase the existing conversion;
- The required number of combinations;
- If you, for example, want to know which title on the website converts better, then this is one combination;
- If you want to change the title, but at the same time change its text and size, then it will be two combinations;
- The number of visitors per day on your site (of course, the average);
- The number of visitors who will take part in the tests.
And that’s it! Press the calculate button and get a period after which you can already draw conclusions about the test results.
Therefore, this is another stone in the “garden” a / b test. With regards to its not relevant to the explosive growth in sales.
HOW TO DO?
Now I will tell you how to do it right. The title of the letter, the call-to-action button, the availability of prices in the commercial offer. All this can be implemented both manually and with the help of special services.
And immediately, without much explanation in love. Manually doing this is forbidden, because time will take a lot.
And if you still have a lot of time. It will be better and more useful for you to implement any channel from the article “Effective advertising channels”.
We use only automated services in order to conduct testing. Usually, when using them in the basic configuration, there is always the opportunity to do tests.
For example, in the designer of the Profit Canvas landing pages https://profits.win/canvas/ you can easily create 2 versions of the page and start testing the landing page without leaving the office.
Google analytics experiments
If we are talking about individual services, then they too. But, as a rule, only for sites. We use experiments from google analytics in our work.
This is a free way and gives us all the necessary analytics.
Therefore, my recommendation is to use it for A / B testing. Here’s how to set everything up – support.google.com/analytics/answer/1745152.
Other A/B Testing
Optimizely.com. One of the most famous services for conducting such studies and tests. Great functionality allows you to conduct the most intricate research.
- Cons – it is paid.
- From the pros – there is a test period and a visual interface (respectively, you do not need to understand the HTML code).
NEED READY LANDING PAGE?
Make a stunning landing page yourself on the Profit Canvas designer.
This is high quality and 100 times cheaper.
GO TO CONSTRUCTOR
If you have a question whether you need to do an a / b test or not, then I’ll say at once that it’s necessary. And this is not even discussed. The only question is when.
And our practice shows that the best option is when the company is already settled and a small change will greatly affect the turnover.
P.S. And one more note that you need to take into account. Do not trust experts who say that this needs to be changed. Because it does not work (according to the sensations). As I wrote above – everything must be tested.