Info Feed Weblog
Dec 16, 2002
Business Automation: When "not yet" is the better answer
VC legend Vinod Khosla has given an interesting speech on creating the real-time enterprise and what to watch out for when automating business processes. This speech has been nicely summarized by Phil Wainewright. Choice quote:
"Anything you do today is legacy next year." This is the philosophical cornerstone of his proposition. Change is inevitable and persistent. Therefore you must design for continuous migration. "This notion of an architecture that accommodates continuous migration is very important," he told delegates.One of the challenges of any business automation is to find the sweet spot between automating too much too early and too little too late. Interestingly, I have found, many managers have a tendency to automate too much too early. Why? Because they assume too early that they have found the perfect solution already, which prompts them to automate around that solution. Finally, because they will soon find out they have been too early in automation (by seeing better solutions, discovering holes in their original solution which had been exposed only after lengthier usage, and most importantly, finding that their problems constantly change, thus never allowing real automation), they'll find themselves continuously updating their automation or, even worse, never using it in the first place.
As a manager, I had always made a point of not automating too early. Example: We had always used an Excel project list for a company-wide overview, prioritization, and high-level progress tracking. That was despite various team members proposing other solutions that promised to be more convenient / faster / less cumbersome / web-based -- or in one word: automated. Yet, I never went away from Excel. Why? It worked for us and any effort in automation would have been very risky unless the advantages would have been paramount (which it wouldn't have been at that time). On top, our environment (both inside and outside the company) was changing so fast that, I found it nearly impossible to gather future requirements for such a solution, much less could I map down the processes to be automated. Staying with the low-tech, unexciting, quite manual, but functioning solution was the right thing. (Almost) Nothing trumps flexibility and ease of use. Especially in an enterprise that tries or is forced to act in real-time.
Posted by Stefan Smalla on Dec 16, 2002 at 17:34 | Permalink