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
Nowadays, the forms of Chinese companies going overseas are becoming more and more diverse. Many companies have established independent websites to directly reach overseas customers. However, the cost of obtaining high-quality traffic will be even higher. When high-quality traffic arrives, improving traffic conversion and user stickiness will bring higher ROI.
Reasons such as mismatched options, inconsistent content, and complicated operations will lower the conversion rate. For example, if a user clicks to view a product he or she is interested in, only to find that the product is not what he or she needs, or cannot find a quick purchase button, this will result in a quick bounce. Therefore, when enterprises serve different audiences, they need to cater to the preferences of different audience groups, and a better solution at this time is AB testing. Take the Volcano Engine AB test platform (DataTester) as an example. The MVT experiment (visualized multi-variant experiment), multi-link experiment (split URL test) and grayscale publishing capabilities provided by it can help enterprises complete the combination strategy of website page elements. Set up, test more accurate conversion forms, improve user experience and reduce churn rate.
Why is the MVT experiment highly suitable for overseas enterprise scenarios?
For website scenarios that are frequently used by overseas companies, usually page optimization is not a complete revision of the entire page, but a refined strategy, such as a certain page element that needs to be deleted, a certain piece of copywriting that needs to be optimized, and a certain button position that needs to be moved. wait. MVT experiments can modify multiple elements and test data without significantly revamping the web page, producing unexpected optimization effects.
The MVT experiment of Volcano Engine DataTester is also called a multi-variant visual experiment (full name: Multi-variate Visual Test). It can start AB experiments on two or more variants of a web page at the same time to compare the combined effect data. It can support more low-cost verification and improvement. Operators without coding background can conduct configuration testing and easily modify product details page copy, shopping cart button, coupon amount, product title, etc., so that these contents can be used as variants for multiple purposes. combinations and form different test versions to ultimately discover a better strategy.
Taking the e-commerce scenario where overseas companies are concentrated as an example, we will introduce the practical application of the Volcano Engine MVT experiment:
As shown in the picture above, on the product details page of an e-commerce website, staff can optimize the design of product display pictures, buttons, and copywriting, and form a combination. So how to use the MVT experiment of Volcano Engine DataTester and how to implement rapid modification testing?
-
Element : Experiment with multiple elements on the page. The variables are product display pictures and copywriting prompting the purchase button to place an order.
-
Variant : After the content or style of an element is modified and saved, it becomes a variant. For example, if the original "Click to Buy" is changed to "Add to Cart", "Get a Coupon to Buy", and "Buy Now", then after the modification is saved, it will be a variant of the action button.
-
Combination : Combination is an experimental group. There are multiple variants under one element. Different variants in the element can form a version by crossing each other. For example, if you are testing 3 Elements, each Element has 2, 3, and 4 variations, and each variation has different elements to modify the content and style, there are a total of 24 combinations (2x3x4).
When overseas companies operate their own websites, whether it is CTA text modification, web page background/button color adjustment, event pages, new product pages, marketing pages, or the upgrade and revision of the entire website, every adjustment can actually be applied to Volcano Engine DataTester to verify the effect.
In addition, there are continuous small adjustments, and there will be major revisions. During a major revision, if you want to gradually distribute some of the existing traffic of the web page to new pages by gradually testing the rhythm of going online, you can choose to use DataTester's multi-link experiment to achieve this by creating a new URL as the new page. Link testing until a new solution is finally selected.
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}}