Call centers

Greg Costikyan tried to buy his daughter a camera phone. Because of complexities associated with multiple account holders, different account types, SIM cards, the AT&T Wireless/Cingular merger etc., it ended up taking hours and the eventual solution was to "[establish a] "new" account, and what they'll do is ship me the "free" phone that comes with a new account. When I get it, I can take the SIM, stick it in Vicky's 3200, and use the free phone as a hockey puck or doorstop or something. Oh-kay. Whatever. This seems remarkably wasteful, but so long as it solves my problem. And hey, it saves me $25."

At the end of the post, Greg nails it in one:

The whole thing reinforces to me that customer service outsourcing is a false economy, particularly if your internal systems are so complicated and your rules so hidebound that your customer service people practically need a college degree in Navigating Our Company's Idiotic Backend. The Indians I talked to were completely useless, I consumed probably a year's worth of profits on Vicky's account in hours of support from US-based CS people, and the Indians in the mix merely made the whole thing take that much longer.

The problem is not so much that the Indians cannot do a good job, it's that you really do need a masters degree in Navigating Our Company's Idiotic Backend. The reason why the backend is idiotic is some combination of 1) being a mix of systems that were never integrated successfully and 2) system needs to support a spaghetti of business rules that are both complex and poorly defined, and nothing kills system functionality like complex, fuzzy rules.

Fixing these has no immediate business benefit, because you spend some money and get no new functionality. All you really get is a lower cost of making subsequent changes. Instead of the basic ROI calculations used to justify business expenditures, maybe something with real options that highlights how the ability to do things more cheaply and with lower risk in the future would help justify budgets for "re-factoring" initiatives. But you know that once you've mentioned real options, you have already gone off some deep end.