Opening the book…
A qualitative assumption hides its own consequences; a number exposes them. When you commit to "the wire is about 1 mm thick" rather than "thin," you can propagate that value and see whether it dominates or vanishes in the result. Numbers also make assumptions falsifiable: a stated figure can be checked against reality, while a vague adjective can always be reinterpreted after the fact to fit whatever you found.
When you catch yourself writing "small," "fast," or "negligible," stop and assign a figure with a unit, even a rough one. Write it explicitly at the top of the problem so it can be challenged later. Then carry it through the algebra and check at the end whether the answer is sensitive to it. If halving the number barely moves the result, you were right to neglect it; if it swings the answer, your guess just became the crux.
Some quantities are genuinely unknown or irreducibly variable, and a single number misleads. There, state a range or a distribution rather than a false point value, and track how that spread propagates into the answer.