We are currently creating a new test homepage that we will divert part of the traffic from the main site to so that we can test conversion rates.

This site unfortunately will not have all the pages the original site has so they proposed sticking links in the top navbar to the missing content. Is there a best practice to this sort of thing?

Reading some other questions on here I came across this solution: Linking a Navigation menu item to external URL

would this work?

Would this solution work for an end user? Because we are redirecting the user to the original site, is that even an "external" link?

3 Answers 3


It depends if the test homepage is created on the same domain. If it is not, then yes, the links would be considered external links.

I would either create a call-to-action style homepage that only offers a user a few unique options that directs them to specific pages on the original site, then monitor the user flow through analytics to get an idea of where most users are being directed OR I would include the same header/menu on the new homepage to give it the same feel as the orginal site and split test it with google analytics experiments.

With Google Analytics Experiments. You are able to test different versions of homepages on a specified percentage of incoming traffic. I've used this before and it is extremely fast to set up and easy to use


If I understand you correctly, you wish to do multivariate (A/B) testing on parts of your site.

This is much more of an implementation (programming) affair than a UX one. From a UX point of view, users shall be completely oblivious to the fact they see a site different to that of others.

Redirecting users to an external site is typically counter-business (you don't want users to leave your site). But as you said, they don't really leave the site, they merely navigate to the A site pages from the B site pages.

There are multitude of ways to achieve what you are after, but that's a question to post to StackOverflow.

From a UX perspective, whether users see site A or site B (which is really part B part A) they should not be aware of it.

  • Thats one part I am unsure of. I guess to your point this is indeed multivariate testing. But I guess typically that is testing minor elements such as a module being slightly different between A and B. In this scenario - our A looks completely different than B. Apr 10, 2014 at 18:48

I completely agree with Izhaki's answer for your particular case. Since the site you're redirecting your users to is supposed to be the "same" as the original with some minor variations, those links shouldn't be considered as "external".

Concerning your question from a general point of view, I wouldn't put any external link in the main navbar. Navbars are meant to help navigation within the same website (as stated here).

The user needs to navigate among sections of a website and while having clear indication of what section the user is currently looking at.

When the user clicks on a navbar links, she usually expects to navigate to a section of the same website, so it would be kind of shocking to land on an external site.

Links should provide the user a reasonable amount of information scent in order for her to feel safe while navigating your application, so external links should be presented in a way that clearly shows what they are. Which way depends on the way information is organized in your application, here's an interesting article from Nielsen Norman Group.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.