For context: I work designing one-off science exhibits. Generally we design around pretty sloppy tolerances when we can, so things can be cheaper to make, with less precise tools, and at more facilities. We sometimes make everything in house.
Question: Occasionally, we'll call out a GD&T feature for a vendor, on a welded frame for instance. Parallelism, lets say. I generally don't expect the vendor to actually check this parallelism directly, but use it as a way to indicate design intent: what's important, and what's not. Should the one-off part come back significantly off, we will escalate to actually measuring the feature. We can then use this feature spec deviation to talk the vendor about a replacement.
Is this an appropriate use of GD+T? Is this CYA (Cover Your Ass) type usage relatively common? How often are GD+T features called out but not measured by the fabricator, or our QA until a problem arises?
Certainly our usage case is not the most common one, and of course for aerospace and etc, the answers will (potentially) be different.
Just wanting some outside voices of what the correct approach is.