Actually, in the case of Fender and other company's brands that I have designed for, we will generally start looking at specific target performance parameters to filter what approaches are likely to be better candidates so that we can focus our efforts on driver designs that are likely to meet our needs. Cost is not generally considered (within reason of course) when initially developing a speaker. Then, we will look if there are any existing designs that are close, with a little luck we may have a driver in our sample inventory that wasn't "perfect" for a previous project but deserves another look. We might also take a design that is close and tweak it for a particular application, but it's not too often that we are able to use a truly off the shelf stock part.
In the case of the drivers used in the higher powered Rumble amps, these were based on drivers previously used by one of their brands, with known performance and reliability parameters. It's less risky this way, as well as working with a more common, unified theme. The same applies to the Bassman Pro neos, as well as some of the SWR speakers used with the GoLite cabinets. This is where the economies of scale benefit comes from acquiring multiple brands, unfortunately those brands are no longer in production.
I have found that Fender in general makes pretty good choices for their speakers, and prioritize the quality over simply the lowest cost. It's tough to beat them at their own game by replacing a speaker with something pulled from off the shelf unless spending quite a bit more money than the replacement.