While this may seem like cause for a long yawn, there's a bit more to it. We should be alarmed when we see insipid rules or sloppy thinking in prose, and this guidance aims to reduce it. While the guidance is specific to drug labeling, some of the thought capital for indications for use can probably be generalized to devices, diagnostic tests, and so on.
The guidance also discusses what limitations and contraindications are reasonable and helpful to add to labeling and which are not. It urges that clinical trial results be on-point and described concisely and clearly.
Devices often have both intended uses and indications for use, and the differences can be cryptic or irregular. But in the simplest explanation, the child's version with training wheels, an intended use is what the device does and often from the perspective of the clinician - device to surgically excise prostate tissue. The indication for use is a typical patient - patient with bladder obstruction requiring reduction in prostate tissue. 510k decisions generally hinge on the broader utility, the "intended use" such as "surgically excise tissue."
Not directly related to this guidance, the FDA also tries to avoid "pseudospecificity" - for example, a drug approved to treat depression in people with blue eyes. Yes, you could do a trial only in people with blue eyes, so it meets the "trial design" test, but it wouldn't pass a "sniff test" or "common sense" test. You might have an antifungal approved to treat fungal infections of the foot, but you can't have "the only antifungal approved to treat infections of the left foot." The resulting label would be "pseudospecific" and, like a half-baked patent idea, it would hopefully be caught and nixed before publication. However, this also shows that you can't do regulatory science without judgement, as there's no simple written rule that tells you if a label or trial design veers into pseudospecificity rather than a legitimate targeted population.