Note: This file has moved to notablog.

I asked: > > What should the "best" e-commerce system contain? Michael A. Stone writes: > first and foremost, an e-commerce system should be based on a very clear > understanding of how the organization in question uses communication as > part of its business. that's the basis of your system design, and the > thing which will lead you to other, more specific questions. Words to live by for any system design, but I guess I was assuming more of a narrow case, the kind of situation most of us here would typically deal with (though, come to think of it, if measured by the dollar, my past experience is more the opposite, with large corporations). Perhaps that points out a need to break it down a bit more: By type: Retail Business-to-Business By size: Billboard web site with single-product, single-click ordering small-to-medium sized company large company For type, we can assume that vertical backoffice (inventory / POS / shipping / order processing / accounting) solutions fill a separate but related niche (or set of niches). Otherwise the discussion gets far too large and is about software design in general, instead of web software design and e-commerce design. For size, use the fairly conventional definitions based on orders of magnitude; a small company employs tens of people, medium hundreds, large thousands, with accompanying budgets. Obviously, in some cases, mainly large companies and companies that specialize in web business (like Amazon, Egghead, etc) the back office will be completely custom-built and integrated with the e-commerce technology, put together by huge projects involving hundreds of thousands to millions of dollars of hardware investment and as much in development costs. Billboard sites, at the other end of the extreme, are fairly straightforward. While they could be said to provide "e-commerce", there's not much more to building and running them than a conventional form and making sure you have SSL so payment data isn't exposed. That leaves the small-to-medium businesses, where the site is hosted on a third party server or on a server dedicated to supporting a particular e-commerce solution. (or, rarely, on a colocated small scale server, like a Linux box, or behind a company's office T1), and interoperability with the backoffice systems is minimal or non-existent, so far (*). (* and likely to stay that way. I'm not familiar with any market-dominating, low-cost, generic backoffice technology. Until the custom developers in this arena start working with more mainstream interoperable technologies, connections to their systems will have to be custom developed to fit. The best that can be done now is to make sure the systems you're developing are ready to play nicely with others; ODBC compliant, maybe someday CORBA support for the big players, etc.) > too many organizations go charging off after the buzzword toy of the month, Too true; I guess I'm trying to get beyond the buzzword syndrome. Or maybe I'm trying to hijack the buzzword process and make E-commerce mean something substantial, so I can use it to talk to people who care about buzzwords. Maybe that's the most constructive thing we can do about the problem. > > What should it cost? > > no single answer is possible, for all the reasons above. Having narrowed down the range a bit, now what should it cost? Or more to the point, if a client comes to you with such a project, what should you budget for software (either bought or for the time to find a GPL package and install and configure it), what should you budget for custom software development, and what should you budget for doing the site itself? I remember _The Geek's Guide to Internet Success_, by Bob Schmidt if I recall correctly (also a member of this forum - hi Bob!) suggested that a web site should cost $5K and hence you should limit your target audience to companies that make $20 million/year. He arrived at that number by applying a rule of thumb about gross profits vs. amount spent on marketing and assuming that the customer is only going to blow part of their marketing budget on a web site. What should the numbers be for e-commerce? > the one thing i can say, with absolute confidence, is that a good > solution.. not overwhelming, just functional and reliable.. will cost a lot > more than the customer thought going in. wish-lists are big, quality > costs money, there's never enough time, and training goes slowly. Ain't it always the case... a good book about *that* is Ed Yourdon's recent _Deathmarch_. And of course there are Steve McConnell's books, _Code Complete_, _Rapid Development_ and _Software Engineering_ (not terribly impressed with this last, but I understand it's more of a cookbook for engineers who don't have time to read the theory in _Code Complete_; _Rapid Development_, on the other hand, is more management focused).