12 revolutionary tips on conducting A / B tests

The beginning of the year is an excellent reason for change. In the first days of January, entrepreneurs often start new projects and try to bring existing ones to the next level. There are many ways to increase the effectiveness of online projects. These include A / B testing, which increases the conversion rate of the resource. In this article you will find 12 revolutionary tips for conducting split tests that will make 2014 a success for your business.

Tip number 1: use the scientific method

Split testing in internet marketing has a lot to do with scientific research. In particular, pharmacists often investigate the effectiveness of new drugs, combining volunteers into two groups - experimental and control. To ensure that the A / B test results are valid, conduct it in accordance with the scientific approach.

The following guidelines will help you conduct split testing using the scientific method:

  • Ask a question denoting a problem. For example, ask why the conversion rate of your online store is 0.5%, when this figure for competitors reaches 1.5%.
  • Investigate the indicated problem. Use Google Analytics or other web analytics tools to determine the entry and exit points of visitors, the time spent on the site and other behaviors. At this stage, you must understand how the audience behaves, and also suggest the reasons for this behavior.
  • Formulate a hypothesis. Based on the data obtained in the previous step, assume how you can achieve the desired result. Suppose with Google Analytics you have found that most customers read the description of the goods, but do not go to the order page. In this case, the hypothesis can be formulated as follows: the conversion rate will double if you increase the size of the conversion button and highlight it in red.
  • Test the hypothesis experimentally. Perform an A / B test, in which half of the visitors are directed to a new version of the landing page with a large red conversion button.
  • Evaluate the results of the experiment. Having collected enough data, estimate whether your hypothesis was confirmed. If after increasing the size of the conversion button and highlighting its color conversion increased from 0.5% to 1% or more, the hypothesis is true. If the conversion remained the same or fell, the hypothesis was not confirmed. Unlike scientists, internet marketers can take liberties when evaluating the results of an experiment. If the conversion rate increased from 0.5% to 0.8%, consider that the hypothesis was confirmed, even though the conversion increased by less than two times. Perform a series of additional experiments to achieve the desired result.
  • Record the results of the experiment, draw conclusions and announce them to your colleagues.

If the hypothesis is confirmed during the A / B test, you can implement the appropriate changes and move on to new experiments.

Tip # 2: collect enough data

What should be the sample size when conducting split testing? The answer to this question depends on the specifics of your project. The main characteristics required to determine the duration of the experiment are the average monthly site traffic and the cost of your product. However, even here there are no specific guidelines, so you will have to rely on regularities.

First, the higher the average attendance of a resource, the longer the experiment should be. If the audience of your online cosmetics store is 50,000 unique visitors per month, several thousand people should take part in the testing.

Secondly, the more expensive your product and the less conversions you get, the fewer visitors you need to conduct an experiment. For example, if you sell expensive industrial equipment and make no more than 100 transactions per year, a few hundred visitors or 10 conversions are enough to evaluate the experiment.

Tip number 3: cut off the noise

Noise is any atypical factors that distort the course of your experiment. Imagine the following situation: you start a split test and talk about it in a guest article for a popular blog. His audience is interested in your experiment and follows the link to your site. Because of this, within a few days the traffic to your site increases by 90%.

In such conditions, your A / B experiment will not bring a valid result if blog readers do not belong to the core of your target audience. Transitions from the corresponding blog are noise distorting the results of split testing. Therefore, they must be cut off. Most of the tools for conducting A / B tests allow not to take into account when testing visitors from selected sources.

Tip # 4: Get Feedback

Study and consider the opinions of your customers by planning A / B testing. For example, if you create a new slogan and plan to test several options, first get feedback from users. Ask how they evaluate your product, what tasks they solve with it, etc. This will give you a lot of ideas and allow you to create slogans that meet the expectations of customers.

HubSpot conducted an experiment in which members of the target group made changes to the landing page. The conversion rate has really increased, albeit slightly.

Tip # 5: Test Usability

Usability testing is one of the best search tools for problems that your website visitors face. So try to actively use it.

For example, if you want to improve the efficiency of an online store, then try to find and eliminate the factors that prevent visitors from searching for the right products and placing orders. To do this, set specific tasks, such as finding and buying a specific product. Using the experiment, determine how visitors are looking for this product, what key phrases and widgets they use, whether they are not distracted by tooltips or offers to subscribe to the newsletter, etc.

If you determine that tooltips really prevent users from placing an order, do additional testing. Ask the audience to use the site after removing the obstacle and check the conversion dynamics.

Tip number 6: use the site search as a source of ideas

Search the site tells you the intentions and needs of users. If visitors are constantly typing in the search box the same words, for example, “price” or “price”, consider whether it is time to “hang” the relevant content in the most prominent place.

Also, searching the site allows you to make sure that you speak the same language with users. Imagine you are selling software. If every second user enters the search phrase "programs to clean the computer," perhaps you should change the name of the section "System Optimization Software"?

Tip # 7: Optimize the most popular conversion paths.

Use Google Analytics to determine the paths your customers use most often on the site. This will help determine the sections of user routes where potential customers most often stop and leave the site.

After receiving this data, you can correct the situation. This can be done by adding calls to action, removing unnecessary menu options, redesigning pages, etc.

Tip number 8: do not be afraid to turn everything upside down

Performing a split test, marketers are often limited to changing the color and size of the conversion button, rewriting the title or text. Such changes are easily performed and immediately bring results.

However, if you really want to bring the project to a new level of work, try larger-scale innovations (for inspiration, we highly recommend reading this article). Just do not rush to completely change the design of the page or introduce several new elements at once. In this case, you just do not understand what exactly brings you success or failure.

Instead, conduct a split test that significantly changes the behavior of the audience. For example, increase the free trial period for your product or remove product category pages from an online store site. Such experiments not only affect the conversion, but also allow you to determine the additional needs and preferences of the audience.

Tip # 9: Use the Pareto Principle

According to Pareto's law, 20% of the effort brings you 80% of the result. This can be used to increase conversion with A / B tests.

Imagine that your site consists of 10 pages. Two of them get 80% of the traffic. Focus on these pages. Do as many split tests as you need to reach the target conversion level on these pages. Only then proceed to the optimization of the following pages.

Advice №10: check several hypotheses at the same time

A standard A / B test involves testing a simple hypothesis. For example, you can assume that changing the color of the conversion button will increase the conversion rate. To test this hypothesis, you send traffic to the base and experimental pages.

You can significantly save time by testing several hypotheses at the same time. For example, try increasing the conversion with a new button color, changing the location of the conversion form, various call-to-action options, etc. To do this, you will have to create several experimental pages and divide the traffic between them and the base page.

Tip number 11: check the response to price changes

Product price is one of the most important factors affecting conversion. Every business sooner or later faces the need to raise prices. However, price increases almost always lead to a temporary reduction in sales.

To test how your audience responds to price increases, do an idle A / B experiment. Create a product test page with new prices on it. However, when going to the checkout page, the user must see the old price of the product. If the number of additions to the cart on the experimental page does not differ from that on the base page, feel free to raise prices.

Tip number 12: do not copy someone else's experience

Copying someone else's experience, even very successful, is the worst way to conduct split tests. Your business is different from business colleagues and competitors, even if you use the same business model and ordered the design of the site in one studio.

Changing the color of the conversion button can increase the conversion of the site "Messenger" or "Euroset." And you can fight for a whole year trying different colors, but you will not achieve statistically significant changes. The fact is that each hypothesis is formulated after studying the problem, which was discussed in the first council. Users may not notice your conversion button, not because of the color, but because of the poor layout. What is the difference, what color is the button if the visitor has to scroll through three screens to see it?

How to succeed by conducting a split test

Implementing split testing, try to independently find serious problems and formulate hypotheses. Remember, standard A / B tests are similar to fast food. Hamburger will save you from hunger, but its constant use is fraught with serious consequences.

Use the tips described above to diversify approaches to page optimization and to find non-standard hypotheses. Thanks to this, you can constantly improve the efficiency of your project.

If you have any questions on A / B testing or you just want to get a free consultation from our specialists, please call (our managers will be happy to talk to you) or send an e-mail to [email protected]

Watch the video: Increase Your CRO by 65%+ Through These AB Testing Methods. AWasia 2018 (March 2020).


Leave Your Comment