Is AB testing overkill for proven usabilty enhancements?

Hi everyone,

Today I was advising my company that our various sites registration forms should have form labels above each field as opposed to using only in-field labels.

I explained that research has been conducted by many people to support this and cited the nielson groups page on the topic I even have had a user mention this in a recent user test. The user said that once the field was filled in it was hard to see what that field was during validation errors. And when distracted users have to return to fill in again and cannot tell what the field is.

So I advised labels above each field.

The reply I got from the marketing director was happy to test.

My question is, knowing that form labels are good idea to improve usabilty, should we AB test?

I know I could conduct another usability test but should we be doing either of these and eating into the limited UX budget when there is already so much research available to support the case for labels not being in-field only.

Is there a line you cross where you don’t AB test every small usabilty improvement or should you infact test every small detail? Even previously proven ones?


I’d say no – the research has already been done. Unless stakeholders are arguing and you need to prove your point, I’d use the budget elsewhere.

1 Like

Congratulations. Yours is the only time I have seen anyone suggest AB testing in a case where it would be valid (that is: a single element with only 2 possible options) However. I have to agree that there’s no developmental or design purpose in doing so. The research is done and definitive; doing it again is a waste of time and money. There IS a rhetorical reason for doing it: You may need to convince stakeholders and the research isn’t enough. That’s the only case in which I’d do the test.

1 Like

My former job was an ecommerce site. According to the widely recognised wisdom, orange buttons had the best click through (I don’t know if that is still true). We tested that alongside various other colours and funnily enough, it wasn’t true for us. Green was the best colour for us.

I believe the latest thinking is that the specific colour is of little to no relevance. It’s the von Restorff effect that you’re after. If you had a predominantly brown/red/orange site then an orange button isn’t going to convert well. If, on the other hand, you have a blue site, it likely will.

If you prescribe to the general theories of colour psychology, orange is all about stimulation, energy and anxiety. Some brands leverage that for things like “Buy Now” buttons, claiming that its just the right amount of anxiety to make you click the button. I’m not sure I agree.