How to determine the value of design

Often times, elements in a design are relatively easy to implement (such as changing the color of a button, or updating an error message).

However, there are other times when implementing a new design principle will involve a lot of coding and time (such as redoing the header of the website)

The time spent improving design costs money, but in order to know whether spending that money is worth it, there should be some method of determining the value (in dollars) of the improved design.

There are only two methods I can think of:

  1. Assume that all good designs are worth it, and spend the time doing it. If a design is taking too much time to implement, then the code was not well written, and needed refactoring.

  2. Guesstimate based on how useful you think the new design will be. You could even perform some A/B testing for more accurate results. (However, getting statistically relevant results from A/B testing may be expensive as well)

I personally think that #2 is the better option, but I'm looking for ways that companies determine the value of a design.