Test. How well do you understand internet marketing metrics

“Don't take anything for granted. Only solutions from practical results, that is, test results. Everything should be tested. I do this all the time.” (c) Gary Helbert.

And in this I completely agree with the King of selling texts (in other words, the most famous copywriter of our time).

A/B testing is exactly what is needed to make the right decision, and not just based on your own feelings. What is it and how to conduct it, we will talk in this article.

distant distances

By default, A/B testing comes from English. Where else it is called split-testing or split-testing. But the most interesting thing is the translation.

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 purpose of the actions is to find out which group of elements will have a higher conversion or indicators.

That is, everything is quite simple. You create 2 different versions of promotional products (let them be emails selling your services) and at the end of these emails you do a different .

And just consider the conversion of which letter was higher. This is exactly what an A/B test is all about.

Here is a good example at the site level, where we determine how the changed block arrangement affects the result. Namely, the lineup raised up.

A/B test

But if you are testing, for example, 2 landing pages in which the headings, call-to-action buttons of different shapes and sizes are changed, then this is not just an a / b test, but a multivariate testing or A / B / N-test.

So why is there so much talk about this test now? Everything is very simple. Marketers like to show their expertise through all sorts of tricks, which can be attributed to testing.

And company owners dream of finding a magic pill and believe that split testing is just a panacea for their promotional materials (in particular with).

What will this test give you?

In fact, not everything is so bad and split testing of the site really benefits companies, and can change the situation in a short period of time, namely:

. The simplest and at the same time the most desired indicator of any owner, especially in websites.
Changing behavioral factors. Not the most obvious, but again a factor influencing the increase in sales.
Increase in the average check. This is the addition of a different kind of call to action or, using the example of sites, the familiar block “buy with this product”.

In short, it all comes down to increasing sales. And it's definitely worth your attention.

But there is one "BUT". And perhaps at this stage you will realize that you do not need it.

This “BUT” suggests that achieving explosive sales by replacing one element, if possible, is very rare.

Approximate chances are 1 to 1000. Since there is a replacement of small components, which rarely differ dramatically from each other.

And when this is a drop in the ocean for you, it is better to focus your attention on more important components.

In addition, while you do not have established traffic, while you are only testing different channels and methods of advertising on the Internet, you may not even start testing.

Since the indicators will not be correct, because the traffic is not homogeneous, but, as you know, different people act differently.

If you have no experience in conducting A/B tests (otherwise you would not be here), then I highly recommend that you read the case studies first.

And especially pay attention to the cases of people who have achieved an increase in conversion / behavior change in Runet. So you will understand what is better to test.

Or you can go the other way. Make a preliminary list (plan) of what you will test on the site.

This should be done based on your assumptions, feedback from the focus group (worst case), or based on data from . This is so that you are not thrown from side to side.

Important. Forget multivariance and only do one test at a time. In our experience, a bunch works best - 1 test = 1 change.

Otherwise, you will not understand what actually gave the result. And it will be even more interesting when one change gives +0.5, and another change on the same page -0.5 to the conversion.

Which results in 0 and an overlooked positive impact element. So don't make that mistake.

Important. Forget vague theories. From the series “something needs to be changed in the site, making it more selling”.

The normal theory for testing is that the button is red on one page and blue on the other.

On one page there is a price, on the other there are no prices. Only then will you be sure of the results. Clear criteria = clear results.

Important. A result obtained in a very short time = poor result.

Especially if in a short time your sites were visited by at most 20 people.

Therefore, before you rush into testing the 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 don't get bored, I found a calculator, thanks to which you can calculate the optimal duration for testing the site - vwo.com/ab-split-test-duration/

Calculator

It is in English, so just in case, I translated and deciphered the lines that you will need to fill out:

Conversion of your page at the moment. How to count, read here -.
The percentage by which you want to increase an existing conversion.
The required number of combinations. If you, for example, want to know which title on the site converts better, then this is one combination.
If you want to change the heading, but at the same time change its text and size, then these 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.

Therefore, this is another stone in the “garden” of the a / b test, with regards to its irrelevance for explosive sales growth.

WE ARE ALREADY MORE THAN 29,000 people.
TURN ON

How to do?

Now I'll tell you how to do it right. The headline of the letter, the call to action button, the availability of prices in , all this can be implemented both manually and with the help of special services.

And immediately, without long declarations of love, it is forbidden to do this manually, because it will take a lot of time.

And if you still have a lot of time, then it will be better and more useful for you to implement any channel from the article.

“Creator of the website about increasing conversion - Boosta.ru, head of A / B testing service Changeagain.me
A/B testing has been one of the main trends in online marketing abroad for 5 years now. Proper A / B testing allows you to increase conversion, income at the same cost to attract customers.

Why high traffic does not produce the desired results? Why do users come to the site, but do not take the necessary action? What design will be the most attractive to users? It is difficult to find answers to these questions in theoretical materials. After all, what works in theory does not always work in practice. A/B testing is one of the testing methods based on user reactions.

Why use A/B testing?

Attracting customers every year becomes more costly. If before the competition was weak and each new client "cost" d e Chevle, now everything has changed dramatically.

In almost every area we are seeing "tough" competition. And those who are guided by the old approach to Internet marketing are constantly increasing their budgets for customer acquisition in order to stay afloat. Of course, you can't do without it. But you need to spend money on Internet marketing wisely.

And for this, abroad, large, medium, and even small companies on the Internet have been using A/B testing.

Why?

Because after you have brought a potential client to the site, it needs to be “processed” correctly. But it is almost impossible to find the right approach to the client without A/B experiments.

If you have created a site with a certain structure, headings, texts and other elements, then you cannot know whether it works as efficiently as possible or not.

Perhaps by changing the texts, making them concise, more understandable, you can increase the effectiveness of the site? Or will changing the color of the button increase its clickability?

What is A/B testing?

This is the process of testing two versions of the same page, where the first version is the original version and the second version has some changes (for example, the title, the color of the call to action button, or its text has been changed).

During testing, all traffic is divided 50/50, as a result of which some users see the original page of the site, and the rest see the test page (with changes). In addition, when creating an experiment, a goal is selected (for example, the number of clicks on the Call to Action Button or the number of applications sent), based on which the most effective variant will be determined

After the experiment you get exact numbers about which version of the page has the best conversion. So all your site improvement decisions are based on hard numbers, not personal opinions and assumptions.

This is the point A/B testing.

At first glance, this may seem very simple, especially if you usespecial services for testing. We wrote a detailed review about them on the site. Boost. en . However, not all so simple.

Conditions for conducting A / B testing

To spend successful A/Bexperiment, the following minimum conditions must be met:

100 conversions per experiment variant: original and test
The experiment must be run for at least 7 days, as user behavior can be very different on Sunday and Monday.
Statistical significance must exceed 95%.
Traffic should be roughly uniform throughout the experiment

As you can see, not every site can conduct such testing. In order for the experiment not to be delayed, it is necessary to have about 10 conversions per day to keep within 3 weeks. In this case, already part of the sites, in principle, cannot apply A/B testing in your work.

As for 7 days, this condition is very important, as many "specialists" in the field of increasing conversions like to pause the experiment in advance, seeing a serious difference in current results. But by no means should this be done. The experiment must run for more than 7 days.

If you plan to use special services, then the third condition will be met by itself. For example, if you do an experiment with Google Analytics or Changeagain. me , then an automatic algorithm works there, which terminates the experiment when it reaches statistical significance.

Traffic homogeneity also plays an important role in A/B testing. Imagine that you attract all visitors to the site with the help of a certain advertising campaign. And suddenly you decided to publish an article on Zuckerberg Call, after which a large number of visitors poured onto the site, who for the most part are not your target audience. In this case, it will be "wrong" traffic that will spoil the results. A/B test.

We advise you to listen to these 4 conditions, since in practice many people very often sin with this, and then complain that A/B testing is completely ineffective. Everything depends on you!

So now you know what A/B testing and how to do it correctly. Surely, you have a question: “What to test?”. There are many options, which we will discuss next.

What can and should be tested?

Call to action button ( CTA -button). You can change its color, text, size, location, shape, number of buttons.
Text elements. Every website has headings, subheadings, descriptions, so you can experiment with them. Test different words, text size, text tone (positive/negative, active/passive), reviews, product descriptions, etc.
Visual Elements. You can test different images on the site, their size, location. When it comes to videos, you can experiment with different types of videos, their duration, voices, scripts, automatic / manual inclusion
Forms. You can test the number of fields in the form, the layout of the form, divide the form into several stages, add explanatory text, various bonuses, time limits
Social buttons. You can test the location of social buttons on your site, their format, color, text that encourages clicking on them.
Navigation. You can change the sections of the menu, remove unnecessary and distracting elements from there, or vice versa, select an element, add icons there, or even remove the menu altogether. In addition, you can test the placement of the menu: side, top, bottom.
Pages with prices. In this case, you can test how the allocation of one tariff will affect the number of subscriptions. You can experiment with the prices themselves: instead of 100, set 99. You can also add stimulating bonuses, gifts, discounts, the ability to use the product for free for several days, give additional guarantees and write about the possibility of a refund within 30 days, etc.

And this is just a small number of general ideas for A/B testing. In fact, there are many more. Especially for this, we even wrote an e-book "100 ideas for A/B testing”, which we send to all subscribers of our site Boost. ru.

However, just take an idea and launch a successful A/B experiment is unlikely to succeed. The fact is that you need to look for problem areas specifically on your site. To do this, you need to use services for web analytics (for example, Google Analytics and Yandex.Metrika), services for analyzing user behavior (GetGoodRank), services for conducting surveys.

By collecting quantitative and qualitative data about your site, you can formulate a good hypothesis for A/B testing.

A Good Example for A/B Testing

Your site has a very small conversion of a visitor to a registered user. This is problem. Using the Form Analytics function of the Yandex.Metrica service, you analyzed how users fill out the registration form on your site. You see that many stop and do not fill in one field that is not the most important. Based on this, you formulate a hypothesis that if you remove this field from the form, the conversion to registration will increase.

Then you make a test variant in which you remove this field, run A/B experiment, wait a while, get the results and analyze them. If A/B The experiment was successful, congratulations! You can move on and look for other problem areas. If the experiment was unsuccessful, do not be discouraged. Look for other problems and start a new experiment.

Some statistics about conversion and A / B testing

Now you know the basics A/B testing. In principle, you can even run your own experiment. And if you still doubt that A/B testing is very effective, here are a few numbers that speak for themselves:

For every $92 spent on traffic, only $1 is spent on conversions. Thus, everyone invests in increasing the flow of visitors, but few care about their future fate. Focus on increasing conversions and A/B testing, and you will already be head and shoulders ahead of the competition.
The average conversion rate in an online store is 3%. On the information site - 10%. Do you want to be average? Continue to be. Do you want to be better? Constantly spend A/B experiment and improve your site.
Successful companies run 50% more tests than their weaker competitors.
A/B testing during Obama's presidential campaign brought him an additional $60,000,000 in donations.
85% of online marketers in the US and Europe plan to focus on increasing conversions this year.
Almost 50% of digital marketing directors already consider the process of increasing conversions to be a key element in the success of their companies.

Good facts, right? You know what's even more convincing?

Imagine that your competitor is testing their site right now and trying to increase their conversion rate. If he succeeds, then from the same amount of traffic he will receive more income, which he can invest in growing his business, fighting you and attracting more potential customers. So don't delay!

For quality material A/B testing thanks to the site

Advertising activities, as well as marketing in general, are considered very creative professions. And all creative people, as you understand, are against restrictions, evaluations and objective judgments (how can you evaluate creativity with the help of tests?). Nevertheless, there are still certain standard and basic knowledge that a specialist in these areas should also have.

If we are talking about Internet marketing, then these are, of course, tests for knowledge of Google AdWords advertising tools and the Google Analytics analytics system.
They are located in the Google Testing Center.
I will tell you about them in more detail.

Google Advertising Knowledge Tests

Despite the fact that at first glance the AdWords interface may seem scary, complex and incomprehensible, once you work in it for at least a month, the picture appears, and many things seem logical and even obvious. It is noteworthy that in addition to the advertising tools themselves, there are a lot of additional features, such as a conversion optimizer, testing ad spaces, working with sitelinks, keyword suggestion, retargeting, linking to Analytics, and much more.

Therefore, after working for a while, I decided that I would easily pass this test. The test itself is 120 questions. It takes two hours to complete. To successfully pass the test, you need to score 80%. You can submit in Russian.
The only caveat is that in order to gain access to this test, you must first pass the Google Ads Fundamentals test. Similarly, in two hours you need to answer 120 questions, gaining a minimum of 85%.

So, armed with practical knowledge and self-confidence, I answered all questions and scored 71%. Ambush, test failed. Re-testing only after 2 weeks is possible. And by the way, EVERY test needs to be bought for 50 dollars (for us, an understandable way is only a card).

And after that, I decided that I still need to prepare. There is a good Google help for this (https://support.google.com/partners/answer/3154326?hl=ru&ref_topic=3203989). In addition, there are good video tutorials on AdWords.

In this case, you can fully view, read and listen to everything in Russian and successfully pass the test in Russian as well. This time I successfully scored 91% and received such a nice certificate:

I want to draw your attention to the fact that I passed the test specifically for search advertising. In addition, there is also a test for knowledge of media advertising, but it is already in English (these are banners, video ads, etc.). This topic doesn't really interest me.

As for the questions, here are a few so that you can evaluate their level:

What is the best algorithm for optimizing display ad campaigns?

Which line of ads will be disabled due to non-compliance with Google's advertising policies?

Each question has 3-5 answer options, from which you need to choose one, the most correct one. Although there are questions where you need to choose several options. As you can see from the questions, even though the test was for search, there are also for media.
The peculiarity is that SEVERAL answers can be correct for some questions, and you need to choose the MOST correct one ...

But, in general, if you read the help and have real experience with the system, it’s not difficult to pass. I passed both tests (the basics of advertising and search advertising) in one breath, for three hours in total.
I recommend it to everyone who works with this tool.

Google Analytics Individual Qualification

This test is more difficult, although more useful, in my opinion. Its major drawback is that it is entirely in English. It costs the same - $ 50, but no preliminary tests are needed to pass it. Bought and give to yourself.

There are 70 questions in the test, 90 minutes are given for answers. The passing score is 80% (increased, before it was 75%).

And in two weeks I scored 78%. Here is a bummer. Slightly missed 80%. Well, besides, because before the passing score was only 75%, i.e. technically, I passed.
But, in general, it almost does not count. Therefore, an English-language forum was added to the Russian-language resource http://productforums.google.com/forum/#!overview

As a result, the milestone was passed (87%) and the certificate was received:

Well, a little about the questions themselves. I quote them in the original, in English. By the way, the questions change by about 30% each time (but the most difficult ones, to which I did not know the answer or was not sure how to evil, were repeated).

How would you set up Google Analytics to attribute conversions to the original referring campaign?
– Conversions are attributed to the original referring to the campaign by default
– Add “utm_nooverride=1″ to the end of your campaign URLs (correct)
– Add “utm_nooverride=off” to the end of your campaign URLs
– Conversions are attributed to the most recent campaign and there is no way to modify thadwordsschoolis behavior

Your website uses a shopping car on a different domain. To ensure that visitor session data is ¿related? when ¿satching? domains, which of the following do you need to implement?
– the _setLocalRemoteServer() method
– The _trackTrans() method
– the _link()or_linkByPost()method (correct)
– This is tracked automatically, no additional configuration is required

What technologies does Google Analytics use to track visitors?
– Java and four temporary cookies
– JavaScript and third-party cookies
– JavaScript and first-party cookies (correct)
– HTML only

Which of the following may only be added to a profile by an administrator?
– Custom Alerts
– Annotations
– Goals (correct)
– Advanced Segments

Why would anyone want to create virtual pageviews?
– To track session ids
– To track a website that uses frames
– To track a website that uses style sheets
– To track visitor activity that does not generate a pageview (correct)

Well, here are some questions. I immediately provided answers so that you roughly understand the format. I want to note that some of the questions (about 15% are technical, about virtual actions). A lot about conversions. There are questions about multichannel sequences. Of course, about evaluating the effectiveness of advertising, including the connection between AdWords and Analytics.

The fundamental difference of this test is that during it you can pause for an indefinite period. The main thing is that the test be completed within 120 hours. Those. You can write difficult questions for yourself and then look for the answer. But, alas, there are no direct and clear answers to many questions.

To successfully pass this test, you really need to understand Google Analytics, know where everything is, what reports, how to build and understand the principles of its work. And most importantly - do not rush, as I did. First, read everything, digest, write out important points or bookmark them. Answer all the questions, take a break, and in a day review the questions again (especially the doubtful ones, it is possible to put marks on the questions and return to the marked ones in the future).

But in general, there is nothing unrealistic in these tests. You just need to really prepare, and not hope for luck. In this case, it won't work.

Choose the correct answer

ordinary consumer

In marketing, you play the role of a silent extra. Yes, you are one of the millions of consumers whose behaviors marketers study.

You are ready to begin to understand the topic, but, unfortunately, it is very difficult for you to concentrate on one thing. Probably, you decided to take the test just for the sake of interest, as you are always interested in something new. And if now you dream of becoming a marketer, then tomorrow you will want to take courses in profiling, Japanese, or culinary master classes. It's never too late to start. Most importantly, clearly define your goals!

Marketing Specialist

You are fluent in terminology and understand metrics. You are well aware of the technical component of an effective promotion, and knowledge is probably backed up by practice. But do not slow down the pace of professional growth!

You know your business very well. Your work is always done with high quality and on time, you are confident in your success. However, you lack zeal. Try to get out of your own comfort zone, look through several blogs, analyze case studies, and you will understand that there are other approaches to doing your job. You need an incentive to improve yourself!

Professional Marketer

You, like a fish in water, are familiar with the terminology and methods of web analytics. Such vast knowledge is certainly backed up by vast experience. Employees consider you a true professional and an expert in your field, and therefore listen to your recommendations with pleasure.

You put your heart and soul into every action, constantly read specialized literature, and sometimes even create your own methods and approaches to data analysis. And all your colleagues recognize your level of knowledge and authority. Experiment with new metrics, train colleagues - this will allow you to hone your skills even more. Congratulations! You are already one step away from becoming the best of the best!

Exceptional genius

To be honest, we did not think that someone would pass the test with such a result, so we did not come up with a description ...

In fact, your results are amazing. You either peeped the definitions of terms, or you are a perfectionist. It is very important for you that all your affairs are brought to perfection, moreover, you have a unique talent for finding answers to any questions.

You are a real fan of your business, you can easily think over a strategy for promoting a product in a few minutes, decide on the tools. Each action is thought out several steps ahead, and vast experience allows us to reduce the likelihood of errors to zero. By making adjustments to an existing project, you bring it ever closer to triumph. Your colleagues are lucky to work with you, because such a marketing guru has a lot to learn.