Recreating a beautiful screen can improve visual observation. It does not automatically improve your ability to choose a flow, interpret research or resolve a trade-off. Practice helps when the exercise matches the skill you are trying to develop.
View full image
If every exercise ends with another polished screen and no evaluation, it becomes difficult to tell whether you are improving or getting faster at familiar work.
Keep one constraint stable
Suppose you want to improve hierarchy. Use the same booking content in two layouts: service, description, duration and price. Change the grouping and emphasis, then ask someone what they would compare before choosing.
Keeping the content stable matters. Otherwise, the version with a shorter description or more appealing photograph may win for reasons unrelated to the hierarchy you meant to examine.
For interaction design, choose a different exercise: take a simple selection flow and add an unavailable option, a changed selection and a failed request. These states demand decisions that a successful first attempt would leave invisible.
Give feedback somewhere to go
“Version B looks better” is a starting reaction. Ask what became easier to find and what became less noticeable. If you emphasized price, did people miss an important condition? The benefit and the cost belong in the same evaluation.
You do not need to turn every exercise into a formal study. A short, clearly scoped review can expose a useful issue. Just avoid presenting a colleague's preference as proof of customer behavior.
Save the decision, not only the image
Keep a brief record of the question, the alternatives, what you observed and what you would change next. Over several exercises, that record can reveal repeated weaknesses: perhaps you handle the default state well but neglect recovery, or improve contrast while making the hierarchy too uniform.
Use the next exercise to investigate that weakness. If the current work no longer asks you to make unfamiliar decisions, another variation may offer little value.
Software makes these experiments easier, but the important loop is choosing a question, producing an answer and exposing it to feedback. The exercise earns its time when the next decision becomes more informed, even if the final screen never enters your portfolio.