Ok, makes more sense now. So I think it's the way you've built/structured navigation then is causing issues, not A/B testing. Then it might be worth fixing navigation to support A/B testing rather than developing your own A/B testing?
I still don't understand what are you trying to A/B test? If you can provide more details it would be easier to help you. Basically, what variant A means to you, what variant B and what is the conversion you are measuring to understand the winner?
Something like this. Easy case (A/B test just one page):
- Variant A: display default hero banner on the homepage featuring generic most popular product
- Variant B: display personalized hero banner on the homepage featuring most popular product in specific country and personalized by country
- Conversion: purchase completed
- Thesis to test: creating country-personalized hero banner gives higher conversion rate into purchases
More complicated case (A/B test journeys of multiple pages):
- Variant A: checkout wizard containing multiple steps (pages) with small forms A, B, C, D
- Variant B: checkout wizard containing one large form (page) E
- Conversion: purchase completed
- Thesis to test: multiple smaller forms give you better conversion compared to one long form
And so on. Both scenarios are very much doable, I implemented those personally. Stock A/B testing works, it must be something wrong with your setup. For instance, it requires Visitor cookies enabled. If you disable cookies - A/B testing will not work.