> For the complete documentation index, see [llms.txt](https://because.gitbook.io/because/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://because.gitbook.io/because/a-b-testing/how-split-ratios-affect-your-a-b-test-results.md).

# How Split Ratios Affect Your A/B Test Results

### **What is a Split Ratio in A/B Testing?**

The split ratio determines how traffic (or impressions) is divided between your test variants. For example:

* **50/50 Split:** Each variant gets an equal share of impressions.
* **70/30 Split:** One variant gets 70% of impressions, while the other receives 30%.

### **How Does the Split Ratio Impact Your Test Results?**

1. **Balanced Split (50/50):**
   * **Best for Accuracy:** Both variants receive equal traffic, minimizing variability and maximizing statistical power.
   * **Quick to Detect Small Differences:** Because both groups have the same sample size, it’s easier to detect subtle changes in performance.
2. **Uneven Split (70/30):**
   * **Favors the Dominant Variant:** More impressions go to the higher-priority group, making it a good choice if you’re leaning toward one variant but still want to test another.
   * **Slightly Less Efficient:** The smaller group (30%) may need more impressions or time to produce reliable data, especially for detecting small differences.

### **Does an Uneven Split Affect Statistical Significance?**

No matter the split ratio, the threshold for statistical significance (e.g., p < 0.05) remains the same. However:

* A **50/50 split** is generally the most statistically powerful option, as it ensures balanced and consistent data across both groups.
* A **70/30 split** may slightly reduce the efficiency of your test because the smaller group has fewer impressions, which can increase variability.

### **When to Use an Uneven Split**

A 70/30 split can be useful when:

* **You Have a Strong Favorite:** If you believe one variant will outperform the other, allocating more impressions to the dominant variant can be strategic.
* **Business Priorities Favor One Variant:** For example, if you’re testing a new dynamic message but want the majority of customers to see the control version for consistency.

### **Tips for Running A/B Tests with Uneven Splits**

1. **Increase Total Sample Size:**

   To compensate for the smaller group, aim to gather more total impressions. This ensures the 30% group has enough data for accurate results.
2. **Focus on Effect Size:**

   Uneven splits are better suited for larger differences in performance. If you’re testing subtle changes, consider a 50/50 split instead.
3. **Plan for Test Duration:**

   With a 70/30 split, the smaller group takes longer to reach statistical significance. Be prepared to run the test for a slightly longer period.
4. **Leverage Tools for Uneven Tests:**

   Use Because’s dynamic testing capabilities to track performance and ensure both groups are measured accurately, even with an uneven split.

#### **Need Help? Let’s Optimize Together!**

If you’re unsure which split ratio is best for your goals, we’re here to help. Reach out to our support team at [**support@trybecause.com**](mailto:support@trybecause.com) or explore our A/B testing best practices for tips on maximizing your results.