Scientific theories are a matter of constructing models through experimentation, which then predict future experimental results. By this definition, UX is a science, not an art. That’s why good UX people can improve specific details on purpose, and non-UX people are just guessing.
Something that is often overlooked by companies and designers alike is that every time you use a UX process to design something, you should get results of some kind.
Those results tell you what worked and what didn’t.
By scouring the web you will also find hundreds of other examples of things working and not working. A/B tests. Eye tracking. Conversion rates. Analytics.
Add those to your mental toolkit.
But we should not take these as definitive answers. The fact that a red link won an A/B test in one case, doesn’t mean it will win in all cases. When you find a case of a blue link or a white link winning another A/B test you must update your model to explain that.
Results are clues about a larger model of the way people interact with things, and the more of the model you understand, the more you can predict the way people will interact with new things in the future. Things you’ve never seen before.
That’s the point of the UX profession: to design smarter things.
Now, at this point it is important to state that this model does not live in your intuition. You are not the model. Your intuition is what the model should replace.
Common Sense is not a model. It is the absence of a model.
If you make UX decisions by having a meeting with your colleagues, you’re not doing science. You’re doing opinions.
To do science you must follow the scientific method. Develop a hypothesis. Research and design an experiment (your solution). Run the experiment (launch). Measure it. Interpret the data to see what worked at what didn’t.
You will notice that this idea is very different from a “creative process”. There is no phase where you look for inspiration on the internet. There is no phase where you gather reference images and put together a mood board or develop a creative concept.
That’s because UX is not art. It’s science. You’re trying to achieve control of your variables, and predictability of your results. You’re trying to eliminate the effects of randomness and chance. You’re being diagnostic, so when you fail, you have more information to use in your next try.
Eventually, your results will get more predictable, because your model and your hypotheses and your research will become more sophisticated. Your experiments will turn out the way you predicted more often than not, and when they don’t, you’ll know why.
That doesn’t mean you stop doing experiments and getting results! It just means you’ll get what you expect more often.
But the important thing about predictions is that you have to state the prediction before the test. If you merely say “I knew it!” after you know the answer, you might as well pull out the Tarot cards.
Science is like golf: the goal is not to win; the goal is to get better.