Friday, March 03, 2006

Operational Efficiency in Tech companies

we had this debate during lunch while I was attending my block at Berkeley for the BC EMBA program. My Cap Markets professor asked how our operations class was going...to which we all avoided the question since it's ridiculously boring and the other professor was standing 5 feet away.

Anyways, we claimed initially that most tech companies don't understand operational efficiency because they haven't had to. I argued though, that this is going to change and has changed. I'm sure companies like eBay have figured out how to most efficiently perform transactions and list "junk" on their website. I can't imagine they haven't done that analysis....have they? As protocols and hardware become standardized at each level, things will become more efficient. Look at how networking has become easy, now if you tell someone how big an organization is, we can scale bandwidth appropriately.I think OS and hardware and web apps will get to that point fairly soon...the standards are already forming. Once that happens, operational efficiency will decide who makes money on it (essentially dropping your marginal cost for your products, allowing you to drop price and gain marketshare).

I know my organization is not efficient on sales operations and has managed some operational efficiency in our back-end operations (we've got 15-20 BILLION queries running through our systems today...btw, we run the .net and .com registry and largest SS7 networks in the US, so that's how we accomplist those feats). I don't think tech companies can actually figure this stuff out though. How do you resolve how many transactions are running through your system? Is it based on web traffic? is it based on users logging in? Is it based on users who bought something? Is it based on page views? All of these must play some role. The last company I worked for (Verizon Wireless) tried to figure this out but they were hopelessly scrambling to keep their infrastructure running and couldn't spend the time or energy to truly develop a holistic view of what actually went through the network. Nobody inside could even figure out what the bigger picture looked other than some of the Technical Directors, but they spent all day firefighting new problems. Amazed me that nobody wanted this information badly enough to spend significant $$ on it!

It got me thinking that it would be a great idea for a company. Wouldn't it be great if someone could tell you exactly what kinds of transactions populate your network. And based on some criteria of what actually encompassed a transaction, you could categorize the different types of transactions as well as model what an increase would look like. Ideally, you could also develop a way to decide on hardware and software that would allow you to understand what your bandwidth needs are.

No comments: