A redesign changes how a product looks. UX work changes what it asks of people and in what order. The two often ship together, which is why they get confused. If the brief never questions a single step in the flow, it is a redesign whatever it is called.

The one with an address. Concentrated drop off at a single step is measurable, isolated and usually cheap, so it pays for the rest of the work. Disagreement about what the product is for is the most important and the slowest, because it needs decisions rather than design.

Yes, and you should. Sit five people down, give each one real task, and say nothing while they attempt it. Usability testing at that scale needs no budget and no software. The discipline is not asking the leading question that rescues them, which is harder than it sounds.

Ask about process rather than portfolio. A good UX design agency will tell you what it found in research, what it changed as a result, and what it killed after the findings contradicted its own idea. If nobody can name a decision that got reversed, nobody was really testing anything.

A single fixed step can show up within weeks, because you are removing an obstacle rather than building a habit. Structural work across a whole product takes longer and is harder to attribute cleanly. Fix one measurable step first, precisely so you have something to point at.