For more technical exchanges and job opportunities, please follow the WeChat official account of ByteDance Data Platform and reply [1] to enter the official communication group
For enterprises, the application of large-scale AB experiments can help enterprises improve decision-making efficiency and reduce trial and error costs. The recent Volcano Engine AB test DataTester has become a The preferred collaboration platform for many companies.
It is understood that DataTester, the volcano engine that has experienced multiple version upgrades, can deeply meet various business scenarios of enterprises. In addition to providing comprehensive AB experiment capabilities, DataTester's intelligent publishing (FeatureFlag) capability can meet the needs of grayscale release of new functions, from Different application scenarios, including AB experiments, full-scale launch, and crowd-directed release, help development, product, and operation and maintenance personnel iterate new features in a low-risk environment, and are also favored by many enterprises.
Smart release (FeatureFlag) is an agile development technology that combines function switches + dynamic configuration + grayscale release + configuration management. It can be closely integrated with AB experiments. AB experiments are used to verify and evaluate the effects of different versions/strategies, while smart releases use To further carry out dynamic configuration, crowd-directed release, full experiment and other configuration distribution. After the AB test, DataTester's original intention of designing the intelligent publishing function is to ensure the efficient and safe iteration of each Feature and to quickly implement refined operations for different users.
In the function development stage, DataTester's intelligent publishing capability can separate function online and function validation, helping enterprises to achieve parallel development of multiple functions; it can also achieve one-stop hosting of functions on different terminals and in different environments, improving development efficiency.
In the effect verification phase, with the ability to intelligently publish, when the AB experiment produces a winning version, the experiment can be directly solidified into Feature, and can be published in grayscale in a robust and complete manner. Later, the AB experiment can also be started by Feature with one click. In the background of intelligent publishing, the product can also globally view which AB experiments have been enabled for each Feature and which versions have been released.
During the function launch phase, smart release provides the ability to schedule releases and gradually expand in grayscale. Product releases can be flexibly brought online and offline. When problems are encountered during the release process, you can quickly roll back to any historical version, reducing Publishing risks.
In the product operation stage, the Volcano Engine DataTester intelligent release can realize group-oriented release, such as displaying different functions according to the characteristics of different groups of people. For example, Android users recommend logging in with QQ, and iOS users recommend logging in with WeChat, achieving refined operations for segmented groups and improving
Personalized experience
for different groups of people .
The Volcano Engine DataTester originated from the long-term accumulation of ByteDance. As of June 2023, Byte has conducted more than 2.4 million AB experiments through DataTester, with more than 4,000 new experiments being added every day, and more than 50,000 experiments being run simultaneously. DataTester currently serves hundreds of companies, including Midea, Geti, BSH, Kaishu Storytelling, etc., providing scientific decision-making basis for all aspects of business such as user growth, conversion, product iteration, and operational activities. It will mature Our "data-driven growth" experience empowers various industries.
Click to jump to
Volcano Engine A/B Testing
to learn more
{{o.name}}
{{m.name}}