Avito Launches Trisigma: A Cutting-Edge Platform for Automated A/B Testing and Digital Analytics

Avito has introduced a new product tailored for business and development, created by AvitoTech – Trisigma. This platform is designed for automating A/B testing and comprehensive analytics of digital services. Trisigma aims to assist companies in making informed decisions regarding the quality and efficiency of new developments and updates to existing products by leveraging data analysis and experimental results.

Originally developed for internal use, Trisigma was first implemented at Avito. Throughout its usage, we gained valuable insights and realized it was time to share our findings with the market by launching a full-fledged product. Avito plans to extend its presence in the B2B solutions segment by introducing additional tools for optimizing and automating development processes in the near future.

The platform enables automated evaluation of new features’ effectiveness across various audience segments prior to widespread rollout. Companies can conduct hundreds of experiments simultaneously across different platforms and operating systems, as well as among distinct user segments. It offers flexible test configurations and generates detailed reports on the findings.

Trisigma is capable of managing large volumes of data from digital services, calculating and organizing metrics and quality indicators. It integrates seamlessly with clients’ existing analytics systems (such as various databases, including Clickhouse) to retrieve data necessary for metric calculation. Moreover, Trisigma visualizes results in an intuitive and accessible manner, simplifying analysis even for those who are not professional analysts.

The service supports the entire testing cycle—from hypothesis generation and experiment management to report compilation, leaving the conclusion and decision-making to humans. Additionally, Trisigma can handle high loads, allowing hundreds of experiments to run concurrently within a single product, test multiple new features across different audiences, and analyze their impacts on user behavior. The platform is easily scalable and efficiently utilizes computational resources.

Six technical capabilities of Trisigma:

— **Group and weight configuration:** Trisigma supports splitting features into unlimited groups, each with adjustable weights.
— **Feature distribution across all devices:** Applicable for both web and mobile applications.
— **Processing vast numbers of metrics and segments during experiments:** It can process tens of thousands of metric variations seamlessly in a single experiment.
— **Completely transparent experimentation:** Includes change history, comments, descriptions, and responsible personnel.
— **Safe simultaneous execution of multiple experiments:** Layered systems allow for numerous experiments to run concurrently without the risk of conflicts.
— **User targeting:** Features can be distributed only to specific users by conducting experiments with dedicated user lists and analyzing data solely from users who encountered the test feature (those who received Exposure).

The platform enhances analysts’ efficiency, automating tasks and freeing up to 90% of their time for more complex and creative challenges. It allows for an increased number of tests without the need to hire additional staff; the sole limitation is the team’s ability to generate new product ideas quickly. Furthermore, it improves result reliability by eliminating human errors and accelerates product development with faster feature deployment.

Trisigma links the process of testing new features with evaluating their impact on business metrics, helping companies obtain a comprehensive view of their digital product development and avoid losses from unsuccessful updates.

To learn more about Trisigma, visit here, and you can find the platform’s technical documentation at this link.

The need for an A/B testing platform at Avito emerged in 2016 when two major redesigns were conducted without A/B testing prior to their launch. Significant resources were invested in the redesign, and the desire to avoid delaying the release led to the omission of A/B tests, which proved costly—as evidenced by a noticeable drop in customer metrics following the launches.

The year 2017 was termed the year of rapid A/B experiments, as releases began to consistently pass A/B gates. Analysts worked around the clock conducting tests manually. It became evident that A/B analytics was a bottleneck that necessitated the construction of a dedicated platform, thus establishing the demand for such a solution from the product team.

Our key conclusions:

Without an A/B testing platform, the product team was forced into manual labor, heavily reliant on highly skilled analysts who could take a week to conduct a single experiment.

The implementation of Trisigma removed these bottlenecks by automating analytics, allowing the number of experiments to hinge solely on the quantity of ideas and hypotheses the product team could deliver.

The platform encourages horizontal growth within the company by facilitating the formation of new product teams. Analysts using the platform can concurrently execute multiple experiments, numbering in the hundreds each year.

“Avito Tech aims to become a key player in the Russian market for automated A/B testing tools. The deployment of Trisigma at Avito has increased the number of product experiments from several dozen to over 4,000 annually. Our experience shows that employing the A/B testing approach yields an additional increase of key metrics for Avito, such as revenue or the number of customers, ranging from 10% to 20% per year,” states Ivan Guz, managing partner for business development at Avito.

The Trisigma team has opened acceptance for new testing applications. Users already include major companies from the fields of online education, fast food, and retail, some of whom have completed their integration with the platform.

Trisigma is designed to assist businesses of all sizes—from startups to large corporations creating digital products for the mass market. Companies in e-commerce, banking, streaming services, and educational platforms are likely to find the platform especially beneficial.

Three signs that it’s time to consider an A/B platform:

— If the time for “simple” solutions has passed and every change in the product carries risk and uncertainty.
— If you need to make less risky decisions and reduce uncertainty, as A/B testing provides data in a convenient format.
— If you aim to lower the cost of conducting experiments; with an A/B platform, costs are always lower, eliminating risks typically associated with “launching without A/B.”

We estimate that developing a similar platform with basic functionalities independently could cost an average company tens of millions of rubles and take about two years. Additionally, maintaining the platform’s infrastructure and team would require no less than 50 million rubles.

In contrast, Trisigma is much more affordable as it is deployed in the cloud with all support and operational management handled by Avito. It does not require significant technical resources from clients, making it easily scalable. The exact cost of using Trisigma depends on the specific needs of each client, the number of tests conducted, and requests processed. Fill out the application here, and we will provide an accurate estimate.

Another development from AvitoTech is the Emcee tool for mobile auto-testing, which you can learn more about here.