Given the wide diversity of environments in which a company replicant must provide services, a geographically diversified chain is going to adapt more quickly, given all the good ideas from the field.
On the other hand, new chains have to start somewhere, and often it's the mom & pop approach in some difficult new environment, like Portland's, like Coffee People, that gets it right in many dimensions.
"Parallel prevails over serial" is one important maxim in business. What it means is you'll want to get many experiments going in the face of big uncertainties, not knowing in advance which will succeed, but knowing that "one at a time" is slower than "many at once."
However, not all management teams are prepared to support parallel operations. In this sense, management is a lot like an operating system. How good is it, at managing memory? And don't blame management for every memory hog application an end user might want to load, on insufficient hardware. Typically, they'll plan to pay you a pittance and ask for the moon, moan and complain if you don't deliver.
In sum: the franchise many-storefront model is ideally a theater for parallelism, and therefore better at fitting in, provided feedback from the field is taken seriously; any highly evolved chain was likely once a corner grocery, and probably had high level, well-exercised skills thanks to some ornery customers or other challenges.
Remember: some of the best stores actually pay managers to go under cover as ornery customers, in order to test the effectiveness of their own store training and policies, sometimes scouting for new managers in the process.
