Стаття від експерта сервісу ROMI center
In order to get new customers and increase sales, the site needs to be improved gradually. And how do you know what your audience is most interested in? The most obvious is to simply make the necessary changes and wonder what will happen. Ale tse hour and pennies, yakі do not want marnuvati.
Here A / B testing comes to the rescue. At the article on the butts, it is shown that it is the same as a tool, it is correct to brush.
What is A/B testing
A/B test — tse porіvnyannya dvoh variantіv webstorіka. Other name options: split testing – split testing, A / B testing. Split-testing is carried out to reveal the most effective version of the side and increase the conversion rate.
Take part in A/B testing of the site:
- variant A — side without change, “control” variant;
- option B – version, as a rule, with one change.
For more options, the process is called A/B/N-testing.
Classic A/B test stock – testing the side on the site with different buttons to the call to dії.
Place on side A the button with the inscription “Remove the free consultation”, and on the same side B – the button “Remove the cost of the price”.
Just keep in mind that there are more clicks on one of the sides, just write on the buttons, which is the main option, and continue testing other elements of the side: the color of the button, the color of the button, the background of the side.
New quirks for A/B testing
Putting material on the site and forget about it – a nasty strategy. If you want the site to bring you customers and sell it, you will have to update it regularly. And it’s also necessary to constantly reconsider how the content is taken into account and improved yoga, to get the maximum benefit.
AB-testing the site will help you get it right. After testing you have real data in your hands – prove or ask for the fact that you can improve the productivity of the web resource, increase the number of leads, sales and conversions.
What is revered for help split testing :
- Advertising texts. In paid advertising, non-permanent text can bring up the budget. Those same on the site. Why is that relevant your content, the lesser imovirnist, scho to spend up to new interest and pide from the site.
- Calls to dії. It is important that your audience responded to the site: subscribed to the promotion, bought, shared content. Testing the call-to-action to help you know the correct tone of support and competently place the call-to-action buttons on the side of the site. Try different forms and button sizes, text and graphics, to make it better, like elements to bring more conversions.
- Okremi sides. Test not only the elements of the advertising campaign – creative texts, but also the primary side. Analyze background color, content relevance, block placement. So you know for sure that the elements of the site are good for the site, and the ones that work and frustrate the curvature of the side.
- Leaves from rosettes. Revise those, zmist, layout and other elements of the sheet. Divide the audience by navpil, send one and the same sheet with different topics, and then marvel at which topic they most often opened the rosette. Split-testing will help you to determine what your audience reacts to the most, and more widely spread options from time to time.
- Subscription form. In forms for generating leads, everything is important, from the colors of the buttons to the text. Such nuances can seriously affect the reaction of a koristuvach – spend an hour on the oblivion of a purchase or learn to drink from the site. AB-test will help improve the subscription form and increase the CTR – click through rate.
How to conduct A/B testing
The basic steps of A/B testing look like this:
Croc 1. Draw a hypothesis
We formulate our bachelor about one, maximum two changes, how to improve the conversion rate of the party and its productivity.
For example, we want to change the color of the button on the call to red from green to green, or transfer the rules to the rules for turning goods from the footer to the white menu.
In order to conduct a test, at which stage it is necessary to clearly formulate the hypothesis and determine the result, as we will win the result.
Let’s say I’ve changed the color to the point that the button is more likely to hit and click more often. For zbіlshennyam kіlkostі kіlіkіv my zrozumієmo, scho zmіna went to melancholy.
Moving on to the policy of turning in another place to break it down a little less. Koristuvachіv, yakі zapovnyuyut form on tsіy storіntsі, may change.
Krok 2. Create side options
We have a control side A with a red button and a reverse side B – with a green one. The same logic, as if testing the placement of the turnaround policy: side A in the same way, with the help of the bottom menu, side B with the new promotion of the power, on the side.
If you need to protest a lot of changes, do not stop them on one single side of B. If you lead to a swindler – you can’t tell if the changes turned out to be effective. Better break the process: test one change on the back, then a friend.
Croc 3. Divide incoming traffic
For split-testing, traffic is divided equally between option B and exit side A.
To check traffic during the A/B test, to test different split-testing tools. As traffic comes to you through paid advertising or search engine systems, you can speed up web analytics. For example, conduct A / B testing with Google Analytics, as these ads are installed on your site. Google Analytics has the least functionality, more assignments to split-testing of the parties – Google Optimize.
👉 How to conduct A/B testing in Google Optimize
Also, for conducting A / B tests, you can install plugins for CMS, which you need. Axis of kіlka brown silan:
Krok 4. Run the test
We start testing for hours, how much is needed to collect statistically significant test results.
It is important that the validity of the collection of statistics for the skin variant may become less than 7-10 days. Only then can we say with confidence that the withdrawal of data is not the result of a drop in the flow of some external factors.
For example, you test changing the text on the button. Change the number of options “Razrahunok tsіni” to “Free consultation”. At the same time, competitors launch a great advertising campaign on the Internet and win the call-to-action formula “Raise the price for 10 credits.”
Do a short A/B test to see if your two formulas are more effective. It is quite remarkable that the test can show the success of the formula “Razrahunok tsіni”: the large-scale RK of competitors is included in the test result. Alas, it’s good for supermen to take a marketing action, it’s possible, “Free Consultation” can be done more quickly.
Henceforth, it is important to select the correct hour for the A/B test and do not accept the date before the term.
Krok 5. Choose an option, what a win
To make a difference in A/B testing, vimiryuemo the results of both sides. The results are the same metrics that we wanted to start with. For example, the number of subscriptions to the distribution, clicks from the banner or sales.
There are three main options:
- Side A is better than side B: we leave the side on the site.
- Side B is shorter than side A: we leave the experimental side as the main one.
- Side A and side B showed the same results: we left one of the options for the ruling court.
Important points in A/B testing
So that the results of an A / B test can be trusted, the test can be statistically significant and is carried out on a statistically significant selection of core students. It sounds nice, so let’s take a look at the terminology and the meaning of the report.
Statistical significance of the test
The statistical significance of the test is changed, as it is necessary to remember the hour of processing the results. The validity of the proof, as calculated mathematically. Everything above 93-95% is considered normal: if an A/B test has a reliability greater than 93%, the test can be considered successful and valid. You can make changes to the site, which have shown their effectiveness.
In order to unravel the statistical significance, you can win online tools – they will do the whole work for you and see the results in a simple way.
The axis of the kіlka is sent to the calculators of the statistical significance of the tests:
If you don’t want to match the results of A/B testing manually for conversions, clicks, visits and other parameters, it’s easier to speed up with the basic capabilities of Google Analytics. The second option is to combine data on low income from CRM and from advertising accounts in one star for help cutting-edge analytics. You will especially appreciate the help of online store owners, who want to know where the leather copy and the slips are stained as a result of bringing contributions to the site.
Vibirka
Statistically significant selection is a group of respondents, for which we test change. The skin of a person, as if seeing the side of the site, is the target audience, and the vibrating axis is a small part of the target audience. For example, the first thousand vіdvіduvаchіv.
In an ideal world, be it a choice, it will be 100% stagnant for all our traffic. Let’s say that we’re testing the hypothesis for hundreds of people, then let’s say that the conversion rate becomes 10%. If we expand the selection to a thousand people, we will be deprived of those ourselves.
In real life, everything can be wrong. On the first hundred shows, the conversion rate is 10%. And the axis by tens of thousands – a drop of up to 3% and lose the same for the decision of the audience. On the right, in the fact that not only the elements that are being tested are added to the conversion rate, but the anonymous external factors – the advertising campaigns of competitors, about which they spoke more, gender specificity and the hour to finish.
To calculate the number of inputs needed to conduct an A / B test, for the sake of using special calculators. The stench will clear up this figure in a fallow way in the case of a bad conversion and a unique one on the side. For example, axis AB-testing service for DriveBack or a simple tool from a private retailer.
What clients write about us:
There are already a lot of pluses. Stable hosting, competent podtrimka, shvidke that competent response. And even more loyal to the client!
Test hosting 30 days
Briefly about the smut
A/B testing is a tool that will help you to know the most effective options for submitting your site. Krіm tsgogo, you can revise them on the basis of your whole audience, first make the most important changes.
At the end of the article, three short paragraphs about how to correctly conduct A / B testing of the site:
- Test only one hypothesis. One A/B test, one hypothesis.
- Watch the hour of testing and do not complete yoga earlier. The optimal term is day and more.
- Analyze the results of follow-up. Tse is your support for the hour of planning distant deeds.
Leave a note about A/B testing in the comments, we’ll be happy to hear from you!
