Yandex Metrica: How the new “Experiments” work

Content
[hide]
What are experiments and AB testing?
How experiments help
What is important to know before running a test
How to work with experiments
How to launch an experiment in Yandex.Metrica
Creating a Yandex.Metrica counter
Installing the code on the site
Creating an experiment
Metrics
Experiment results
How to know when there is enough data
Checklist
Results

Willday Editorial

Ask a question
Yandex.Metrica is a free web analytics service developed by Yandex. With Metrica, you can track the number of visitors to your site, what sources they came from (for example, search engines, social networks, advertising campaigns), and find out which pages were the most popular. This is useful for determining email data the effectiveness of advertising campaigns and improving user experience.

What are experiments and AB testing?
Yandex Experiments in Metrica is a method that allows you to compare two or more versions of a web page or application and determine which one gives the best results according to pre-defined metrics. The idea is that users are randomly sent to different versions of the page, and their interaction with each version is analyzed and compared.

How experiments help

The service is quite multifunctional and with its help you can significantly improve the company’s performance due to:

Ability to conduct A/B testing;
Evaluation of design and content effectiveness;


Conversion funnel optimization (creating a chain of events that lead to conversion)
Testing on different audience segments;
Experiments to evaluate the effectiveness of your advertising campaigns;
Obtaining statistical data and conclusions.
ab test.png

What is important to know before running a test

To conduct a reliable and statistically significant A/B test, you need to have enough traffic on your site. This will help ensure that the results are representative and statistical conclusions are reliable;
Determine the planned duration of the A/B test, taking into account the traffic volume and the expected change. The test duration should be sufficient to obtain statistically significant results;
Select a method for splitting traffic between options A and B. Yandex.Metrica offers various options, such as even splitting, cookie-based visits, etc. You should choose the most appropriate strategy, taking into account the goals and features of the test;
interface .png

Experiments interface in Yandex.Metrica

experiment.png
How to work with experiments
Think of the Yandex.Metrica experiment as a scientific experiment. Just like in a lab, when we change only one variable, we can accurately measure its impact.

Data accumulation. The more data you have, the more consider your business needs reliable and accurate your results will be. It is recommended to have at least 200 conversions per month to have statistically significant conclusions.

Analyze all results for the same period of time. This will eliminate the influence of seasonality or other factors that may distort the results.

Distribution of impressions, traffic and conversions is also important to consider. For the successful operation of the auto strategy in Yandex.Metrica, it is recommended to have at least ten conversions weekly.

How to launch an experiment in Yandex.Metrica

Authorization is the first step, without chine directory which you won’t be able to work in the service.

authorization.png

After the first step is completed, the transition to the main page occurs automatically.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top