How can I tell which elements on a screen matter most?

It's mostly a strategy question, and only partly a visual one. To know which elements are the important ones, you need to understand what product you are making, for whom, and why. With that knowledge, what's important is clear, and good design is a matter of reflecting it through visual hierarchy. I love when hierarchy debates turn into this conversation, because it forces teams to agree on what the screen is actually for.

So start by asking what someone needs to do and understand on this screen. Each screen has to make its purpose clear: what can people do, and how can they do it? Keep in mind people often arrive from search engines or aggregators looking for something very specific, and they decide in milliseconds whether your page has anything to do with what they care about.

Then sort everything into three buckets: most important, secondarily important, and not really important. That third bucket? Just get rid of it. Once you know the priorities, assign visual weight through contrast. If one thing is red, circular, and big and another is small, square, and gray, guess which one wins.

E-commerce product pages are a good example. The buy button is the action that makes the money, so it's big, high contrast, and placed right next to the product image, since images attract our attention and the eye goes from one to the other. Same goes for actions generally: give the primary one clear visual priority over secondary ones.

Getting the order wrong has real consequences. I've used a circus poster to illustrate this for over twenty years (yes, I'm dating myself). The performers and title dominate, so you know the circus is in town. Reverse the hierarchy and people walk past posters seemingly promoting "$5.50."

Finally, check your work. Just squint. It blurs the design enough to quickly see whether the important elements stand out. If everything still looks equally loud, your eyes end up bouncing between visually equal elements instead of being guided through a priority. Decide what matters first, then make the squint test prove it.