Track experiments with GA4 / GTM

Bonus question : it seems that if you conduct 2 parallel experiments on 2 separate pages, visitor will see the same alternative number of each page.

Example : if I’m affected nab=0 for pageA on the first experiment, I will necessarily get nab=0 for pageB on the other expriment.

Is it correct?

If so, that would be a problem… That would mean you can’t really make 2 independent experiments on different part of your website at the same time.

Hi!

Regarding your first question, I’m afraid our plugin doesn’t provide any specific integrations with third-party analytic tools like Google Analytics. However, if you don’t hide the nab query arg from your tested URLs, you should be able to view in GA when visitors have seen one variant or the other by looking at the tracked URL.

I’ve been looking at the cookies set by the plugin, but it’s not clear how to interpret them.

The cookie nabAlternative has a numeric value that tells our plugin which variant should be loaded in each test. If it’s even, it’ll load variant A — if it’s odd, it’ll load variant B. If you’re wondering what happens if the test has more than two variants, it basically selects one variant or the other looking at the remainder of dividing this value and the number of variants.

Bonus question : it seems that if you conduct 2 parallel experiments on 2 separate pages, visitor will see the same alternative number of each page. Is it correct?

Yes, that’s correct (as you probably could infer from my previous paragraph). We used to have different values per experiment, but we finally opted for an easiest solution that, as far as we could tell, better matched the needs of our users.

If this is an issue for other users, we might revisit this, however.

Thanks for all the clarifications !

The use case for me is that I want to test in parallel 2 separate steps of my funnel.
For example : Landing page (LP) => Cart page (CP) => Checkout

If I understand correctly, I have no way of testing independently 2 versions of the LP and 2 versions of the CP.

Indeed, visitors of LP #A will all be directed to CP #A and visitors of LP #B will all be directed to CP #B.

Which means, I’m not independently testing both the LP and the CP.
It’s actually one test : LP+CP #A VS LP+CP #B.

I think this problem kind of reduce the benefits of the plugin being able to run 2 tests at the same time.

Yup, unfortunately, that’s a current limitation.