Thanks, I appreciate your perspective on the problem, and that you see the complexities and challenge here. I agree that on both sides of your point that perhaps there is an opportunity here to develop software to fit this niche, yet at the same time many businesses need different workflows. That being said, two thoughts occur to me on that.
First, at least in the case of this flower shop and, I'd wager, many other very small businesses (that are nonetheless needing or at least wanting to start using more advanced tools to help them focus on the really important and unique parts of their business), it seems very probably that they could be made to adapt to a certain approach to their business flow, provided it was clear and simple to use. If a static approach provides enough other benefits (e.g. universal data synchronization, workflow automation, etc.), then it can still be embraced despite requiring changes to existing work processes. And I realize in fact that many businesses do adjust how they do things to fit existing solutions when they see that solution as "the" way to do things, for example Quickbooks. While it's a somewhat flexible tool, it still enforces certain ways of doing things (partly for legal reasons, partly just because it has its particular approach to things). I've heard many times from small businesses owners "Well we do that because that's how Quickbooks codes it" or whatever. Probably that rings a bell for you too. Even if it's not Quickbooks, it's *something*. So arbitrary constraints can be adapted to provided sufficient incentive.
The second thought that occurs to me is some kind of easily and fully customizable system might be used by IT consultants to make a custom(ish) solution, but without breaking the bank. If you want a custom solution from any of the major CRM manufacturers, it is usually major money. An open source customizable solution that could be setup by a competent IT person and not require any significant new coding to establish e.g. a particular workflow, or change the names of primary fields, or whatever, could cost small business owners far less. If it was done right, with lots of modular bits and plugins, all of which could still be updated easily through a central update system (think Linux package manager, or Ruby Gems or similar), it might be a very functional solution, allowing reasonable customization by non-coders (still requiring "experts", but at a much lower and less costly level), maintaining ease of upgrade, and getting people the functionality they need.
I think though that the first approach is really the most practical and I'm quite interested in considering it further. I still maintain hope that someone out there has made the defacto "This is the best way to organize your business, and our tool makes doing so very easy so use it!" system and we can just buy it, I can train them, and then they'll be off and running. I'm quite sure if I gave them a system that allowed them to manage all their contacts, accounts, associated tasks and documents, was accessible from the web remotely, from their blackberry, and integrated/synced with their web store and Quickbooks, they'd be happy to take some time to learn how it worked. They might occasionally curse the rigidity of the system, but 90% of the time it'd be making their work faster and easier.