Something better A/B testing
Hello! My name is Basil. My main occupation is to increase the conversion of sites, the key metric in any IT business.
According to statistics, the average conversion rate of sites in a Runet does not exceed 1 percent and this means that a large number of Internet entrepreneurs every day loses a lot of money. These entrepreneurs can be mentally divided into two main types: the first are those who simply sits and looks at your resource is not optimized and does nothing. Unfortunately, they are doomed to be crushed by the time, innovation and competition. The second are those who are trying something to do with it. Usually, they paint the landing pages, experimenting with TV ads and conduct a/B testing. A/B testing is a very necessary tool of modern marketing, it allows you to determine which landing page is most effective for most users. To understand how this method is effective — it is sufficient to recall the sensational story about Obama and 60 million dollars. However, this method has a downside, it is that everyone is different, and if for example statistics show that women are for the most part love cats, it is not necessary that your girlfriend has no allergies to them. That is, if we A/B test showed that a variant of the perfect landing page for 60% of users and we choose it for our advertising company, we automatically lose the remaining 40%, which it does not cling. So, a/B testing is something to do with it. And right now, because we are the hell lose 40% of leads and is not cool at all!
In order not to lose customers, need a system which not only conducts an A/B test after the fact and shows you the results, and the ability to predict which version of a landing page is like your visitor and show him this option. It is a pity that it is not possible, because no one ever really knows what your potential clients in mind, right? Or not?
Although, of course, we can assume that some of your visitors are not watching a presentation video in the workplace, because they are firing bosses, and if you are on the website came from your grandmother, then most likely it is not going to work selling text which has been written for your son. From head climbs a huge number of matches. What if finding these correspondences automatically, if you create an algorithm which finds the similarity in behavior of different users on the website according to various parameters, such as time, geolocation, browser, operating system, UTM mark with the advertising companies? How much do you think would match it found?
I have to say that we did this thing and got an interesting result, it turned out that with a probability of about 80%, the algorithm can predict which version of the landing page is more like a specific user. Yes, we have no data that eating good for Breakfast, also we don't know what's the name of his dog or wife. But even the data available has been enough to automatically find matching. To make our algorithm work better, we have attached to it the so-called system of weights. If the landing page users do not like and they do not commit to this target action, he slowly descends in the rotation and eventually generally not be shown to user or displayed only to the group that really likes him, only when he likes. Again, there is a small problem lies in the fact that to identify the correspondences and correct operation of the system requires some time and appropriate attendance. And the more options of landing pages in rotation, the greater the need of visitors. For each option, preferably at least 500. Unfortunately, if you have an attendance of 200 people a day and 5 variants of landing pages, you are going to go nuts rather than wait until our algorithm is segmenting your users. However, if you have under equal conditions the attendance of 2000+, he will learn for a day and a half. And if you have this traffic with ads, which includes utm labels, the algorithm will create for you magic. However, do not forget that even a magic Genie doesn't grant wishes, if not to RUB his lamp, to put it simply, if you made 5 choices terrible is not usable pages that differ from each other only by the number of unicorns in the footer, do not expect that the algorithm will help you.
In General, the thing was interesting and I think it looks like the future of marketing technology.
I apologize in advance, if you missed any mistakes in the text, if you can write in the PM.
PS I'm the only marketer with disgraphia in this country.
Article based on information from habrahabr.ru
According to statistics, the average conversion rate of sites in a Runet does not exceed 1 percent and this means that a large number of Internet entrepreneurs every day loses a lot of money. These entrepreneurs can be mentally divided into two main types: the first are those who simply sits and looks at your resource is not optimized and does nothing. Unfortunately, they are doomed to be crushed by the time, innovation and competition. The second are those who are trying something to do with it. Usually, they paint the landing pages, experimenting with TV ads and conduct a/B testing. A/B testing is a very necessary tool of modern marketing, it allows you to determine which landing page is most effective for most users. To understand how this method is effective — it is sufficient to recall the sensational story about Obama and 60 million dollars. However, this method has a downside, it is that everyone is different, and if for example statistics show that women are for the most part love cats, it is not necessary that your girlfriend has no allergies to them. That is, if we A/B test showed that a variant of the perfect landing page for 60% of users and we choose it for our advertising company, we automatically lose the remaining 40%, which it does not cling. So, a/B testing is something to do with it. And right now, because we are the hell lose 40% of leads and is not cool at all!
In order not to lose customers, need a system which not only conducts an A/B test after the fact and shows you the results, and the ability to predict which version of a landing page is like your visitor and show him this option. It is a pity that it is not possible, because no one ever really knows what your potential clients in mind, right? Or not?
Although, of course, we can assume that some of your visitors are not watching a presentation video in the workplace, because they are firing bosses, and if you are on the website came from your grandmother, then most likely it is not going to work selling text which has been written for your son. From head climbs a huge number of matches. What if finding these correspondences automatically, if you create an algorithm which finds the similarity in behavior of different users on the website according to various parameters, such as time, geolocation, browser, operating system, UTM mark with the advertising companies? How much do you think would match it found?
I have to say that we did this thing and got an interesting result, it turned out that with a probability of about 80%, the algorithm can predict which version of the landing page is more like a specific user. Yes, we have no data that eating good for Breakfast, also we don't know what's the name of his dog or wife. But even the data available has been enough to automatically find matching. To make our algorithm work better, we have attached to it the so-called system of weights. If the landing page users do not like and they do not commit to this target action, he slowly descends in the rotation and eventually generally not be shown to user or displayed only to the group that really likes him, only when he likes. Again, there is a small problem lies in the fact that to identify the correspondences and correct operation of the system requires some time and appropriate attendance. And the more options of landing pages in rotation, the greater the need of visitors. For each option, preferably at least 500. Unfortunately, if you have an attendance of 200 people a day and 5 variants of landing pages, you are going to go nuts rather than wait until our algorithm is segmenting your users. However, if you have under equal conditions the attendance of 2000+, he will learn for a day and a half. And if you have this traffic with ads, which includes utm labels, the algorithm will create for you magic. However, do not forget that even a magic Genie doesn't grant wishes, if not to RUB his lamp, to put it simply, if you made 5 choices terrible is not usable pages that differ from each other only by the number of unicorns in the footer, do not expect that the algorithm will help you.
In General, the thing was interesting and I think it looks like the future of marketing technology.
I apologize in advance, if you missed any mistakes in the text, if you can write in the PM.
PS I'm the only marketer with disgraphia in this country.
Комментарии
Отправить комментарий