Having to implement an extra serial port in software is a good example of how a cheap processor resulted in big grief. For a big-volume product, it may be economical but not guaranteed to actually work. For upgrading existing products, it may be the only choice. Designing a new product based on such a processor choice is very dubious. In the end, you really do need the full requirements to be able to select a processor. A cheap processor may also require extra hardware on the outside requiring a larger total component cost, or larger PCB or larger current consumption. More external components may affect the factory testing time.

In the end, the cheapest processor is seldom the best choice. You may not need the most expensive choice, but most often something in the middle is the optimum.