Effective A / B test way and its attitude 23 summary

"A / B test"Is used in the construction of a website as a method to judge more excellent layout and design by preparing multiple patterns of images and sentences, replacing them with each other and comparing user's reactions. 23 points to be noticed in order to make maximum use of the A / B test are released.

23 Tips on How to A / B Test Like a Badass - Search Engine Watch (# SEW)

The A / B test was issued in 2012WIRED articlesWith the attention of world attention, various tools have appeared thanks to its dramatic effect and introduced to online shopping sites and others. However, consulting company on website managementTreehouseEric Shiu, who is representative of the company, has said that from the experience he has helped to improve the client's site, even companies that have adopted the A / B test can not make good use of it .

The 23 lessons from Ms. Shi to make effective use of the A / B test for the site administrator who introduced the A / B test on a trend, but has not been improved, is as follows.

01: A / A test

ByRuss Seidel

Shi says that it is important to first check the reliability of the software in the first place before purchasing and installing dedicated software for A / B testing in earnest. there,Under exactly the same conditionsTry A / B testConversion rateTry competing for superiority or inferiority "A / A testIt is recommended that you try out. If there is a dominant difference in the A / A test, we should doubt the reliability of that software.

02: Eliminate "snag"


Potential customers who visit the site are frustrated with unnecessary input forms and overly long pages, for example when viewing pages. If there is a "snag" under this irritation, the user will be dismissed from the page and puzzle, so it is said that an appropriate condition for extracting meaningful data from the A / B test is not prepared. First of all, she confirms that there is no catch on the site, Shiu pointed out that it should do A / B test after losing a snag.

Specifically, it is important that the input form is simple, the step is made less, and the fact that the page is not too long is an important point in eliminating the catch.

03: It does not produce anxiety

ByLuca Rossato

Just like the snag, Mr. Shi tells us that there is "anxiety" as a major cause that the A / B test does not produce accurate results. For example, if the user is required to perform laborious work on the site, if the user can not judge whether or not consideration corresponding to the labor of the work can be obtained, the user has anxiety and distrust, and again the A / B test It seems to have a bad influence on the result.

04: Focus on understandability

ByJosh Semans

Although it tends to pursue "persuasive thing" when selling things and services online, it is said that it is not always the best to always be persuasive according to Shi. It is more important than 'persuasive' rather than persuasive.

Therefore, in order to make it easy to understand, it is important to be conscious of the stage at which stage of the sales process, what can be done, what can be done, why it should be done so that the user can understand, It is meaningful to conduct B test.

05: The user's voice is first

ByGabe McIntyre

The use of the A / B test is also important, but it is important to ask the customer directly about how the customer came to the page, what action was taken afterwards, what was influenced by it, etc. Mr. Siu thinks that. Mr. Shi will sound a warning that he should not only rely on the A / B test without listening to the user's live voice.

06: Pricing

By@ Doug 88888 Here

It is very difficult to grasp the allowable payment for services. It is very important to make a hypothesis of payment tolerance and check this with the A / B test.

However, as customers are concerned about the value obtained by paying the amount, they should not rely on the test results from the beginning, but pricing is done after the thoroughly that "the service is worth the price" Mr. Siu talks that it is important that it is important. It is not blindly to leave it to the A / B test, but firstly it should thoroughly study A / B tests based on the real world skin sensation repeatedly and then finish the A / B test.

07: High pricing test

By401 (K) 2012

If it is a cheap price, the sales amount increases, and if it is a high price it tends to think that it can not sell. However, Siu says that it often happens that the pricing of services to be offered is too cheap. If the price is too low it may be misunderstood that it is a service of reasonable value.

Even if you raise the price, if profits are dramatically improved if the service is accompanied by value, it is worthwhile to conduct the A / B test for "slightly higher pricing". So, Ms. ShiPrice increase of 2%It is recommended to try. With a slight price increase of 2%, it seems that customers can avoid feeling too profitable.

08: SNS button


I often see "Like" buttons and share buttons at the top of many product pages, but Siu thinks it is very important to make effective use of SNS. For example, when Clearasil of skin care products installed a "Like" button and a Tweet button at the top of the page, data showing that page view has improved by 25% has been issued. The button of SNS is worth to try on the A / B test.

09: AdWords location

ByRichard Masoner / Cyclelicious

Landing pageofAdWordsThere may be common sense that traffic increases as the position of the car is higher. But Shiu says that such common sense should be put on the A / B test.

The higher the position of AdWords is, the higher the advertisement cost is set, but lowering the rank may dramatically lower the cost. How is it different between 1st and 2nd place? How about second and third? Surprisingly, if you lower the position of AdWords, the cost effectiveness may not deteriorate. Mr. Shi recommends that you should do A / B tests without regard to common sense about the location of AdWords.

10: I also doubt the royal road

ByInsomnia Cured Here

There are many "royal roads" created by many previous tests in web design. For example, it is better to have a catch copy of the site headline and a catch copy of the text ad. Indeed it is a good choice to follow these royal roads when there is no time to validate. However, if there is time to test, Siu says she should doubt even such common sense, and should actually try it in the A / B test.

11: Try with a smaller experiment

ByKelly taylor

In the absence of time, Ms. Shi thinks that it is also effective to conduct a smaller test than the A / B test. Data can be obtained in 5 minutes as an example of a smaller experimentSilverbackAre listed. Mr. Siu once saw that one mail order site has improved the conversion rate by 13% using the Silverback 5 minute test.

12: Preferential three elements

ByDmitry Dzhus

Ms. Siu, as three factors to increase the conversion rate, relevance · value ·CTA, And we are explaining the importance of improving these factors. Of course, these factors are subjects to be improved and can be said to be the priority for A / B testing.

13: Do not forget the macro viewpoint

ByJohn Carlos Buen

When it is shown by A / B test that the conversion rate changes greatly even with small design differences, it tends to be a microscopic point of view such as detailed design. But Shiu says he should not forget to change boldly from a more macroscopic point of view. For example, it is a good example of an important macro viewpoint to change the sentence itself, not the design.

14: Do not be greedy

ByDavid Goehring

The important thing for doing the A / B test is to focus on specific "one" aspects on the page and make changes and test it. Shi says the software used for the A / B test also has a simple design that can be understood at a glance rather than one that displays complex data analysis results in detail.

15: Importance of headlines

ByDenise Krebs

It is important to know what to appeal to customers, and it is the headline to play the role. In the headline, you should write only the best points and strengths of the service. However, publishers themselves may not understand something that is most grasping the customer's mind in the service's strengths. Shiu says something "biggest uri" to write in the headline should be explored in the A / B test.

A good way to do that is to first write out all the things that seems to be "good" in your service and list it and continue to apply it to the A / B test to adopt the winning river as headline .

16: Also sticking to fine things


Even slight differences in design can cause dramatic change in conversion rate. Therefore, Mr. Shi tells us that it is worth testing A / B no matter how trivial it is. For example, there are companies that tested A / B as to whether or not capital letters should be mixed, such as "www.YourDomain.com" and "www.yourdomain.com" for URLs to be displayed on AdWords, and the result is that URLs with mixed capital letters It seems that the conversion rate was as high as 53%.

17: Challenging bold changes

ByShiv Shankar Menon Palat

If you gradually refine the site by repeating the A / B test, there is a tendency to avoid making major changes. However, Ms. Shi thinks that it is a completely different design that can overcome the best design obtained by improving until then. Sometimes it is also necessary to throw away the best things we have accumulated so far and to apply a completely new design to the A / B test, sometimes we should also try what is taboo.

18: I do not aim for perfection

ByKevin Gill

Incorporating A / B tests may lead to misunderstanding that building a "perfect site" is a goal. However, Shiu says that the purpose of the A / B test is to make a better site.

19: No need for the latest software

ByTaryn Domingos

Every day, new A / B test software has appeared, and there are some that present complex analysis and graphical analysis results. However, Mr. Siu says that the latest software is not the best and speaks that simple and comparatively cheap software is good even if it can only analyze simple.

20: Moderate balance

BySteven Lilley

Often it is not easy for regular people to communicate when refining the site. Mr. Shi used A / B test of the optimized site and the conventional site, suggesting to the client how to make the site with too much information and having a bad appearance to be simpler, It says that the site had better results. It is important to consider a reasonable balance with customer service without being limited only to site optimization techniques.

21: Wait for the results

ByCadillacdeville 2000

The result of the A / B test may take time to obtain black and white, so it sometimes happens that a long time test may be effective. Mr. Siu sometimes saw the winner and the loser, which had been revealed at the beginning, reverse in the middle when the site was refined for a long period of nine months in a project that requires a lot of work That's right. It is said that you should not rush the results of the A / B test.

◆ 22: Do not block escape


In order to raise the conversion of landing page, there is common sense that it is good to exclude the link which can become the user's "escape path" in the page thoroughly ". However, when I performed the A / B test on the page where Scribe removed the link and all the links, the linked page won. This is also an example of the necessity of the A / B test which suspected common sense and the identity, Mr. Siu talks, and the attitude that doubts common sense is important in the A / B test.

23: There is a test only

ByRachel Cobcroft

Finally, what Sihu is giving as a lesson is that there is no goal that this is good for the A / B test. Even if it is possible to select the best present thing, because the shape of the best site will change with the passage of time, I told that it is important to continue the A / B test and continue improving as to whether better products can always be made It is.

in Design, Posted by darkhorse_log