If you suspect a traffic imbalance issue in your A/B test, follow these key steps:
Understand how ABConvert distributes traffic
Go through our traffic imbalance checklist
Take action based on your findings
Let’s go through each step in detail.
Step 1: Understand How ABConvert Distributes Traffic
ABConvert assigns each visitor a random seed (0-99) on their first visit and stores it in their browser. This seed determines their group allocation based on the test’s traffic distribution settings, ensuring consistency for returning visitors.
However, traffic distribution may appear imbalanced due to the following factors:
Filtering conditions – If certain users (e.g., from specific traffic sources) are assigned to a specific test group, distribution may be affected.
Insufficient sample size – Small traffic volumes can naturally lead to short-term imbalances due to randomness.
Changes made during the test – Adjusting test settings mid-test can cause allocation discrepancies.
👉 Short-term fluctuations are normal. If imbalance persists after 250+ views, contact support for assistance.
Learn more about this topic in Understanding traffic distribution discrepancies in tests
Step 2: Traffic Imbalance Checklist
Go through the following checklist to identify potential causes of traffic imbalance.
Are there test group assignment settings enabled?
Are the traffic settings evenly distributed?
Have traffic settings been adjusted during the test period?
SRM (Sample Ratio Mismatch) result
Let's see how we can check on these points.
1. Are there test group assignment settings enabled? (Summary)
Check whether any of these settings are enabled in your test settings:
Display test price based on audience traffic source
Display test price based on cookie value
Conditions for new and returning visitors
👉 If any of these settings are active, traffic will not follow a strict 50/50 split.
2. Are the traffic settings evenly distributed? (Summary)
Go to the Traffic Split section in the test Summary.
Confirm that it is set to 50% / 50% (or another intended ratio).
👉 If the split is not evenly set, the imbalance is expected.
3. Have traffic settings been adjusted during the test period? (Experiment History)
Check the Experiment History to see if any traffic settings were modified during the test.
Changes like adjusting the traffic split mid-test can cause imbalances because new visitors may be allocated differently from earlier ones.
👉 If changes were made, this could explain the imbalance.
4. Check the SRM (Sample Ratio Mismatch) Result
If all previous checks look fine and your have set your traffic to be equally allocated but it still appears imbalanced, check for SRM in Experiment Health Status.
If SRM = "No" → Everything is running normally.
If SRM = "Yes" → This suggests a deeper issue affecting traffic distribution.
What is SRM (Sample Ratio Mismatch)?
Sample Ratio Mismatch (SRM) is a sign that something might be off in your A/B test. It’s an objective calculation that helps ensure experiment reliability. When you run an experiment, you expect the traffic to be split evenly (for example, 50% of users see Version A, and 50% see Version B). If the actual split is significantly different—like 55% vs. 45%—this signals an issue. SRM is not a guess; it’s calculated using a statistical formula.
How is SRM calculated?
SRM is typically detected using a chi-square test, a statistical method that compares the expected and observed traffic distribution. If the difference is too large, the test produces a p-value—a number that tells us how likely it is that the difference happened by chance. If this p-value is very low, it means there’s a problem, and the SRM flag is triggered. Our app automatically checks for SRM, helping you quickly spot issues.
Step 3: Take Action Based on Your Findings
If you've gone through the checklist and suspect a real issue with traffic imbalance:
Review your test settings – Check if any assignment rules, traffic filters, or mid-test adjustments are affecting distribution.
Wait for more traffic – If your test has fewer than 250 visitors, short-term fluctuations are normal. Let the test run longer.
Contact support – If SRM is flagged and no clear cause is found, reach out to our support team for assistance.
Conclusion
Traffic imbalances in A/B testing can be caused by a variety of factors, including filtering conditions, sample size, and tracking issues.
Short-term variations are normal due to randomness.
Filtering conditions or test settings may intentionally cause imbalances.
SRM checks help detect deeper issues in traffic allocation.
By following this guide, you can diagnose and troubleshoot traffic imbalances to ensure the reliability of your A/B test results. If you need further help, don’t hesitate to contact our support team.