Smart Traffic Allocation: Powered by Bandit Algorithm

In this section, we will explain in detail what Smart Traffic Allocation is and how to correctly configure an Experience deployment utilizing Smart Traffic Allocation. Before reading this section, we strongly recommend that you first review the content on A/B Testing.

What's bandit #

This section will focus on explaining the logic behind Smart Traffic Allocation. Understanding Bandit logic will greatly enhance your ability to conceptualize effective Experiences. Of course, if you are already familiar with Smart Traffic Allocation, you can skip directly to the "Operation Instructions" section.

In simple terms: Bandit Smart Traffic Allocation is an intelligent traffic management system. The smart allocation algorithm will, based on your defined conversion goals, ensure traffic is dynamically directed towards the best-performing variation in real-time.

Ptengine's implementation of the Smart Traffic Allocation algorithm automatically decides, based on the conversion status of each variation, whether to primarily leverage the current best-performing variation or to continue exploring other potentially high-performing variations. This process gradually achieves a balance between exploration and exploitation. This approach saves traffic wasted on lower-value variations in traditional A/B tests, while also eliminating the cost of manual traffic adjustments. Furthermore, traffic allocation adjustments become more scientific and objective. Additionally, it is more adaptable to scenarios where new variations are added mid-test; the intelligent algorithm will not completely reset its allocation strategy due to the introduction of a new variation.

How to set up a bandit testing #

1. When editing A/B test or redirection test, select traffic allocation Settings
   

2. Select the Smart allocation option, and then select the conversion target.
   
   The method for selecting Conversion Goals here is similar to A/B tests. Ptengine recommends choosing the most intuitive metric that you want to impact with this Experience (creative). (For example, if the Experience is a "campaign page traffic pop-up", then "Campaign Page Visits" would be a suitable Goal).

3. Finally, click "Confirm" and "Publish Experience" to complete all settings! You will also observe that the traffic allocation is displayed as "Automatic." In the A/B test report interface, you will also notice differences from traditional A/B tests, as the "Improvement" and "Probability of Being Best" metrics will be absent.

Understanding Smart Traffic Allocation: A Visual Explanation

These two diagrams can quickly illustrate the function of Smart Traffic Allocation, as well as the key differences between A/B testing and the Smart Traffic Allocation Bandit algorithm:

❶ Experiment Differences: A/B Testing vs. Bandit

❷ A/B Testing & Bandit: The Power Combo

A/B Testing & Bandit: A Powerful Combo (Referring to the previous section title)

Example Workflow:

• Step 1: A/B Test - Validate Your Core Idea's Value: (e.g., Testing whether placing the "Add to Cart" section at the bottom vs. middle of the page is more effective)
• Step 2: Campaign Variations - Create Multiple Designs Around the Validated Concept: (e.g., different colors/styles for the "Add to Cart" button)
• Step 3: Bandit - Optimize in Real-Time: Use Bandit to dynamically direct traffic to the best-performing design variation during your campaign.

When to Choose: A/B Testing vs. Smart Traffic Allocation

• Use A/B Testing When: Your goal is to invest time and traffic to rigorously validate an operational strategy and leverage the long-term insights from the results for future improvements.
• Use Bandit Smart Traffic Allocation When: You aim to maximize results and value within a specific timeframe (e.g., a short-term promotion lasting 1 week to 1 month) by optimizing across multiple variations.

If you require further clarification, please do not hesitate to contact our Customer Success Manager for a prompt explanation of product features and benefits.