Hi, we have an A/B test that has been running since 8/8/2018.We are allocating 100% of the traffic and splitting 50%/50%. We use A4T.
For some reason the test is excluding most visitors to the two pages.
The Audience we have selected only excludes our company's IP range (which is already excluded from Adobe Analytics).
There are no collisions.
I would appreciate any ideas.
Thanks for the help everyone. It turned out to be a stupid mistake in the Page Delivery conditions!
Are you using at.js or mbox.js for your implementation? A4T only works with at.js 0.9.6 or later, not with mbox.js. Also, have you looked in Target to see what numbers it is reporting? I would check Target to see what it is reporting and see if its more accurate than A4T.
If Target is correct, then most likely this is only a reporting issue, not an issue splitting traffic. In other words, its quite possible that your test is running correctly and splitting traffic correctly, but its just reporting wrong through A4T.
1)Did you try fetching mbox trace of any visitor who is not recorded in the test?
mbox trace generated in the first page load has 'geo' field in it which can tell you the IP and then you can verify if it is included in the test or not?
2) You can also check for your analytics segment in case you are using that in the target test.
May be these checks help.
Hi Ali, Yes you can still run the mbox trace. It will be visible as part of the target response in the network call. Also since you are using A4T have you validated your A4T implementation.
Ensure that the call are fired in the right order Visitor API --> Target --> Analytics. Also make sure the SDID parameter is set in both Target and analytics calls.
We're using A4T which uses at.js not mbox.js.Can we still run an mBox trace?
As for the filter, I removed it four days ago but it hasn't made a difference in the numbers (the IP range is excluded from Analytics as it is)
Hi, As mentioned could you try running a mBox trace on the page where the activity is running to confirm that users outside your internal IP addressed can indeed qualify for the test. Are you using a profile script with a regex condition to filter out your internal IP range? If so it could be that its too restrictive.
Details on running a mbox trace can be found at Troubleshooting Content Delivery
Hi LJ, the numbers match exactly. Still a fraction of visitors to test pages.
Thanks for replying. We're using A4T which uses at.js not mbox.js.