August 20

"You want to be a tree until it's time to be a branch."
Tim Miller, CommonHealth

Two client projects in which I have the role of program manager have given me reason to think about the benefits of off-the-shelf software and hosted services. Few today would consider a wholly custom-built application, but how much should you customize a canned application? And if you have to customize it a lot, is it worth building from scratch? My presumption is "no", and it's rarely worth trying.

First, A Few Memories
The projects bring to mind two memories:

• Twenty some years ago, my dentist decided he didn't want to spend $25,000 for a dental office management system. He knew BASIC. He went out to Radio Shack, bought a TRS computer, and started building one. After six months of work, he realized he had hardly scratched the surface. In the end, he bought the commercial system.
• Ten years ago, during my tenure at DEC, I was aware of an effort to deploy some major system, and the "build vs. buy" battle was raging. Those favoring a certain million-dollar-plus product lost the battle because "it only did about 65% of what we need". The system never went into production.

Two is not much of a sample, but it's representative of my experience. The projects show why.

Project #1
Our client seeks research subjects and has to winnow a list of about 10,000 candidates down to roughly 20 who will actually participate. They have some basic facts about the 10,000 - enough to be able to pick 1500 to focus on with a mailing. From that point on, it's a matter of individual contact with considerable screening. The problem can be solved with good contact management, some workflow, and enough added capability to keep track of studies.

What to choose....

Workflow and contact management are available off-the-shelf, but not in one product that suits the client's budget. The chosen solution is a CRM system (chosen for its contact management features) customized to add the workflow. Roughly half the cost of the project went to the workflow, as did virtually 90% of the acceptance testing and essentially all of the regression testing after bug fixes.

Was it still worth it to have done it as we did? Yes. An off-the-shelf package gave us far more than we could ever develop in a cost effective manner to solve our basic problem. We got reporting, the ability to import and export spreadsheets (allowing us to import 125 spreadsheets of historic data), a time-tested user interface, standardized training, annual upgrades for very little money, a bullet-proof way to handle much higher loads, mail- and fax-merge (both useful in this application), and the ability to share the cost of the server and the server licenses with other users who may need some CRM capability.

As they say in late-nite TV, "but that's not all!". The company now has a tool that will boost its productivity (read: make it more competitive in the market), aid its growth, and enable it to do more for years to come, partly by taking its business process to a new level.
Did they get everything they wanted? No. For starters, the process they follow is "contact centric" - where the focus is on the contact - while the application is "company centric" - where contacts only exist in the context of their employing company. Incompatible? Not if you're flexible enough to say that (a) everyone works for themselves, and (b) we'll ignore companies as used in the application. And another unmet need: The client chooses candidates so as to minimize researcher travel. This would be aided by a proximity tool, but the cost was simply too high to justify in a first implementation. But, if they add that later, they'll buy it. Building that from scratch would be impossible for a company this size.

They were flexible enough, and they got most of what they wanted for much less than the cost of a comparable custom-built application (think: the 80- 20 rule).

Project #2
Our client operates a highly relationship-driven business that requires them to deliver exceptionally high-quality content to a demanding clientele. Published material is derived in part from interview notes, and material goes through a rigorous drafting and review process.

Our envisioned solution is a blend of relationship management and knowledge management (KM), but not in the way it's typically done in today's CRM systems. These systems use knowledge management as is useful for service incident resolution (tech support); problems and fixes are stored and catalogued, and they are searched during subsequent support incidents. And CRM systems' document libraries are typically for canned marketing materials but are not useful for deliverables. We needed to organize and search interview notes, manage document development, and record document dissemination.

Again, there are many ways to skin this cat. Find a product that does both. Find a product that does CRM and has a partnership with a KM vendor. Find a product that does our brand of KM and has a partnership with a CRM vendor. As long as the CRM and KM are integrated, any of these will do.

There are none for under a quarter of a million dollars.

In the client's price range, two vendors had apparently promising solutions. What did we discover in the bidding and evaluation? Neither solution was fully baked. Simply put, the two "sides" were not integrated: In one, users couldn't send managed documents to any of the contacts; in the other, users had no way of recalling who documents were sent to!

We are about to assemble the chosen solution. We chose an extensible CRM product. Off-the-shelf, we'll take contact management, tracking of all manner of interpersonal relationships, integration with Outlook mail and calendars, remote access through terminal services, untethered access with client software and a mini-database you can travel with, reporting, document and database search, opportunity management, standardized training, and the standardized backup and recovery you get with a commercial-grade database. We'll custom build some of the specialized features. And we'll use the supplied search engine, which can search documents inside the CRM system or in an external folder.

What didn't we get? We've given up sophisticated document management capability, so we'll use naming conventions for managing versions (rev 1, rev 2, ...) and approval level (Draft, Edition 1, ...). We'll also manage our documents in a separate workspace, keeping only those documents in the application that have been or could be delivered to my client's clients or marketplace.
The company now has a tool that will boost its productivity, aid its growth, and enable it to do more for years to come. The tool also provides an implied and embedded business process that serves as a foundation for business processes the company hasn't yet built out. Again, the 80-20 rule served us well.

Bigger Observations

• Business process and software application are opposite sides of the same coin. Successful applications are successful in part because they support best practices or business processes that are common and re-usable. If your business process is so specialized that little of it can be supported by off-the-shelf applications, you should ask yourself why.
• If you think you'll custom build and maintain what you need at lower cost than buying, think about this: If the vendor is spreading the development and support cost over, say, 1000 customers, do you think you can build it and support it for one thousandth of the cost? Ok, so you only use about a quarter of the product. Do you think you can do it for 1/250 of the cost? If you ignore the cost of report writing and querying capability, good user interface design, and production-grade reliability, you're deluding yourself.
• As you customize, be careful about over-mechanizing at first launch. Many first conceptions of need - particularly if complex - don't turn out to be right. Keep it simple, gain some experience, and add complexity after you have some more confidence that your business process is right.

All in all, have your vision, but when the time to act arrives, make practical choices that get you most of the way to your vision. Take your best choice and bend with the wind when standing rigid is self-defeating.

Or, as Tim said, "You want to be a tree until it's time to be a branch."