Problem Statement
Sometimes, you may notice that your VWO test reports show different visitor data as compared to another website analytics tool. This article explains the different reasons that cause a discrepancy in the data results.
Resolutions Steps
For JavaScript issues:
-
The JS code you have entered in the test pages is not executing correctly. If there is an error in the JS code, VWO does not count the number of visitors in test variations.
-
There can be a case for VWO editor-based campaigns, where the variation version has a code snippet applied to it which reloads the page, based on some condition. As a result, the network call that records the user visits of these campaigns can fail further leading to visits not getting recorded.
-
If you are using location.reload or location.replace through the VWO code editor directly or indirectly, it may cause an issue in logging visitors for variation containing these method calls.
-
* Directly: The function is called directly in JavaScript code added by using the code editor for any variation.
-
* Indirectly: In case you are appending/moving some elements which contain scripts with these method calls.
-
In case you have tried applying some custom JavaScript changes which further modify VWO campaign cookies through VWO code editor for any variation.
-
It has also been observed that if we have JS code in variation that sends several additional calls and has several setInterval methods, this would further decrease the loading time of the variation. This leads to an increased bounce rate from variation, causing lesser visitors to variation.
For error in variation targeting
Please ensure that you have defined targeting and segmentation for the variations appropriately. Inappropriate segmentation on variations can give you misleading reports on the test.
For Split URL test
The difference in the reports for a Split URL test can be due to a number of factors. For example, a different split URL campaign is running on the control page of another split URL campaign. It will cause log visitor requests for control to get aborted in most cases.
Look out for such errors, and remember the following pointers for uniformity in reports:
-
VWO code should be present on the Variation URL.
-
Make sure that the VWO code is not installed through a tag manager. Also, if the VWO code is installed at the bottom of the page, it can lead to fewer visits being tracked.
-
Enable Cross-Domain Tracking if the Variation URL is on a different domain.
-
The Variation URL you specified in the test should not redirect to another URL. Otherwise, a visit to the variation will not be counted, because the page redirects to another page that has not been defined in the variation URL of the test. This issue can be fixed by adding a redirection by using JS code and adding VWO Sync code before that.
-
If query parameters are defined in both the control and variation URLs, the variation URL needs to have an asterisk at the end to incorporate the query parameters of Control. For example, you may have the following setup for Control and Variation URLs:
-
Control: http://example.com?abc=123
-
Variation: http://example.com?xyz=456
-
In VWO tests, you need to define the URLs, as shown below. An asterisk at the end of the variation URL will take care of the further query parameters which are added when a visitor lands on the variation URL.
-
Control – URL matches: http://example.com?abc=123
-
Variation – URL matches pattern: http://example.com?xyz=456*