We talked to Red Hat, MandrakeSoft, SUSE, VA Linux (now VA Software) Linuxcare (now Levanta), and a bunch of others. None of them was interested. None seemed to understand that small business owners don't really care about the fine point of Linux or Windows or other technical details; that if you offered a system they could buy or lease for a reasonable price that included maintenance and upgrades, you could build a stable, easy-to-run business that could bootstrap itself financially after a modest initial investment.
The barrier we ran into was what I call the "Silly Valley Syndrome," where giant deals are all that interest anyone, and the phrase "exit strategy" is part of too many business discussions. No one was interested in an essentially humble business, serving other humble businesses, that would generate modest profits from a diverse base and provide a decent but unspectacular rate of return for its founders and investors.
Worse, the Linux software people asked, "What about hardware?" and the hardware people asked, "What about software?" These questions proved that we weren't being heard. Our idea, obviously, was to sell or lease both as a single package. We believed (and still believe) that the ideal small business computer system would arrive ready to plug in and run, preloaded with a stock set of applications, with additional application packages available for an additional fee. We wanted to take all worries about computers off the small business manager's shoulders. We saw a service business, not a hardware sales business, even though hardware would be the base on which the service rested. And everything we wanted to do relied on hardware and software that already existed and worked well.
The network is not the computer
We were aware of Oracle's and Sun's attempts to sell thin-client networks, and in our opinion they went about it wrong. To begin with, their offerings were too expensive for the market we wanted to enter, and relied too much on custom, proprietary hardware. And that whole "network is the computer" thing was an obvious washout. We were not surprised that this concept never took off.
When you build a simple network for small businesses, the server is the computer. The printer goes next to the server, so it doesn't need to be any kind of special, network-attached device. A smart business with a decent budget would want to have a backup server next to the main one. All other computers on the network would be the simplest, cheapest, most reliable units we could find, something like the low-cost white box computer Joe made for me in 1999 that is still used daily by my stepdaughter, Alicia. If new units designed specifically as thin-client desktops cost less than minimum-spec commodity boxes, we can use them instead. It doesn't really matter, as long as they are all the same in any given installation.
LTSP (the Linux Terminal Server Project) has licked all the problems we ran into when we built our first proof-of-concept network, which ran Mandrake and flawlessly served as my home system for several years. Our biggest problem was getting sound working on the clients. The second-biggest was remote printing. Both sound and printing have gotten easier in Linux since 1999. Better yet, "packaged" systems built on a uniform hardware base will never have hardware incompatibility problems. You supply printers, scanners, and other peripherals as part of the system. That way you know they'll work with Linux, and your client won't need to think about them at all. He or she will just compute, scan, and print -- trouble-free.
"Trouble-free" is the key
To most people, "The Internet" is like a black box. They connect to it, and all they care about is that it works. They don't worry about routers and nodes and traffic-balancing and all the sweat that goes into keeping the infrastructure running. A small office computer system should be the same way. It should either arrive, pre-tested, in boxes and get plugged in or be plugged in by a local, low-level network wiring contractor. As soon s everything is hooked up, people in the office should be able to sit down and go to work.
Naturally, if new software is involved, such as a conversion from Microsoft Office to OpenOffice.org, you may need to do a bit of video training and possibly some in-person work, but it should be kept light and easy. And once you've done the installation and training, no one in that office should ever need to think about the computer system again, except perhaps to replace a burnt-out monitor or desktop unit eventually.
You configure the server to be administered remotely. This is what makes this system cheaper and easer to maintain than others. And the price can become unbeatable if the company behind it is large enough that thousands of its clients share the cost of maintaining a wide selection software packages.
Since virtually every American office, even one with only a few people in it, has someone who can unplug a non-functioning monitor, desktop client unit, or even a server, assuming the cables and ports are clearly marked, the vendor should never need to appear in person after installation and initial training, and chances are that most customers won't even need an initial visit. Network wiring is simple, a job now done by handymen who also install telephone and cable TV wiring in homes and offices. There is no real need for a technician from OfficeComputerNow (hypothetical name) to come out and install wiring, and with adequate training material and a decent help desk, the in-person training step may not be necessary, either. But even if that training is not needed, it should still be available, as should a local OfficeComputerNow representative who can take care of any problems that come up.
Sales and support people will still be needed
This product will need demonstration. There will probably be a range of options, since no two businesses have exactly the same computing requirements. Some, for example, may need to have a Windows server and several pieces of Windows software available, plus client licenses for that software. This means selling OfficeComputerNow systems will require sales skill and enough product knowledge to make sensible recommendations.
After the sale, even if the system can be maintained remotely, a local support rep -- who may also be the local saleperson -- will make clients feel more comfortable. Technology is wonderful, but people use it, and people like to talk with other people, preferably local people, ideally in person, when they have a question or problem.
I do not believe it is wise for a company working with small, local businesses to achieve "economies of scale" by dehumanizing support either through Web interfaces or by routing all customer questions to an offshore "call center." The place to save money by having a small staff work on behalf of many clients at once is on the technical side; centralized download servers and a centralized update and diagnosis staff who log into clients' computers remotely, plus hardware and software standardization, is how OfficeComputerNow will keep its support costs basement-low. If most contact with the technicians who do the backend work is through local company reps, it doesn't matter if the techs have strong Indian or Chinese accents. But then again, speed of response and the ability to work within clients' cultural milieu may make it better to have the support center located in the same country as most of the clients. And if one support group serves hundreds or even thousands of client companies, the cost per employee within that support group should not be a huge bottom-line factor.
A franchise opportunity
What we're really talking about here is a franchise situation. A central company should supply the hardware, software, and behind-the-scenes tech support, with local individuals or small companies doing the sales and providing post-sale training and in-person support.
When Joe and I tried to get Linux-oriented companies interested in this kind of operation back in the twentieth century, all of them were going to sell direct over the Internet or wanted to have only their own people working with customers. They were not interested in working closely with independent resellers other than hardware OEMs.
Now that the dot-com foolishness is over, and Linux is no longer considered an "upstart operating system" but has become a normal part of the IT landscape, it seems like a good time to bring up the idea of a service-oriented, Linux-based computer system supplier again.
The plug-and-play Linux office system was a good idea in 1998, and it's a better one today. It is a business model that is absolutely in tune with free software principles, and better yet, it offers a chance for local Linux advocates to exercise their zeal in a productive, profitable manner.
If you know any companies or investors you think might be interested in creating a Linux-based service company catering to small businesses, please call this article to their attention. This business plan sketch is not proprietary in any way. I wrote it to encourage others to act on it, not out of any self-enrichment hope. All I ask in return, if you manage to get it going, is to have system number one installed in my home office.