Dynamic allocation process for the A/B tests has been streamlined
❤️We listened to your feedback and simplified the process for dynamic allocation.
Users had to decide whether they wanted to run a Predictive test or a classic A/B test and make the selection in the user interface. Customer feedback helped us realize we could streamline the workflow to set up an A/B test with dynamic allocation. We listened and have enhanced the whole process to include dynamic allocation directly from the traffic allocation phase of creating an A/B test.
🔍 How does the Dynamic Allocation work?
Dynamic Allocation allows users to set up an A/B test with minimum risk thanks to the Multi-Armed-Bandit algorithm, Thompson Sampling and Bayesian statistical models algorithm. During a test the algorithm will detect the best performing variation and gradually allocate more traffic to it, maximizing your conversions since day 1. Dynamic Allocation can be based on any type of KPI (action tracking, page goal, transactional goal) excluding browsing metrics.
You can find more information about how Dynamic Allocation work here in this article.
💡 Why using Dynamic Allocation ?
Without Dynamic Allocation, users who are running a time-sensitive test and don’t have the luxury of waiting to perform a full standard A/B test would have to manually stop the campaign, reassign the traffic to the winning variation, then finally restart the campaign. This is not the recommended best practice as this is time consuming: users will face the risk of making mistakes in the traffic allocation and the campaign's update as well as risking to lose revenue.
Thanks to the Dynamic Allocation option when allocating traffic, users who are short on time will have the opportunity to directly allocate the traffic to the best variation without having to interfere with the campaign. This will secure quicker results & better ROI!
Dynamic Allocation can be very useful in the following cases:
- When users want to optimize micro conversions that are expected to occur in a short period after the user has been exposed to a variation. On e-commerce websites, for instance, some clients prefer the “add to cart” CTA over the transaction event as the primary goal.
- When users have a very short time to run a test:
- Running promotions for the holiday season with a few variations of the promotional engagement, where the business goal is maximizing revenue during this short period of time.
- Covid crisis pushes a business to communicate with customers, this communication could present itself in a few different variations but the company just wants the winning variation to be used ASAP.
- When a page that needs to be tested has really low traffic. Reaching statistical significance being a challenge due to the low traffic doesn’t mean you shouldn’t optimize the experience. Dynamic Allocation would be a logical choice where the more traffic receives the optimization even before reaching statistical significance.
- When there are a lot of variations to test (more than 6), dynamic allocation enables users to quickly identify the least performing variations in order to run the test on the most relevant ones.
🚀 How to get started?
To use the Dynamic Allocation feature, first you have to create an A/B test. Go to Goals step and assign a primary goal to your test, as long as it is not a browsing metric. Then, on the Traffic allocation step, just click on the "Change to dynamic allocation" button to successfully switch your standard allocation to dynamic allocation. This is the workflow enhancement we hope you all enjoy!
To be able to see the result of an A/B test with dynamic allocation, you can go to the AB Tasty report page or click on the report button on the dashboard page like a regular A/B Test.
Dynamic Allocation is only available for customers in the Growth or Elite package. If you would like to explore upgrading to enjoy Dynamic Allocation and other AI-powered capabilities such as Engagement Level and Content Interest you can contact your KAM or CSM.
If you need more information you can check the product documentation.