Most IT organization spend between 70 and 80% of their resources on keeping things running. Maintenance for servers, patching, network circuits, and solving problems fall under this. Unfortunately that means the amount of IT resources working on anything new is at best 30%. So out of every dollar you spend on IT you get 30% new value.
We have been able to reduce our IT operations costs to between 30 and 40%. I can't give you an exact number because, well, spending the time to accurately track every thing we do would change the numbers. Frankly figuring out if it is 33.8% of 37.4% isn't time well spend and too many departments measure too much and was resources on that instead of new value.
In fact that's my first tip. Stop measuring what doesn't matter. If at the end of the day you aren't going to do anything with the data, stop spending time measuring it.
The second tip is sort of related to the first one. If you can't remember or explain why you are doing something, stop doing it. This can be an outdated process, or a contract for a service that you have had for a long time. We used to have a contract with a data cleansing company. It was expensive, and when we asked, we realized no one used it. So we called in the vendor and asked why we should keep it.
The vendor went on to explain how clean data would save us money when we did postal mail campaigns. We never do mail campaigns and asked what other value they had. As it turned out, that was all they could come up with, so we cancelled the service. The only impact was we saved a few hundred thousand dollars of operating costs a year. No one said anything at all about the service going away because it was never used.
I know this goes against the outsourcing trend, but if it is a lot cheaper to do it yourself, hire someone really good, cross train someone else and bring it in house. A lot of companies buy all of their network services from one of the big telephone companies. The thought it, consolidating to one vendor gets you economies of scale and letting them manage it is one less headache.
While letting someone else deal with it is less headaches, we saved millions of dollars by switching from a managed network to one where we get the best local deals on Internet access we can and manage our own VPN network. We get great pricing, more flexibility, and slightly - very slightly - more headaches. If you hire the right people, the downside is minimal and the upside is huge.
Automate as much as you can. This may mean taking chances that the automation won't miss much. We used to run a management package that would let us deploy updates to our machines. We would test and test each one before we released it to the users to make sure there were no issues. We decided to stop doing that and trust the vendor that makes the software to test the fixes. It's been a year and we have had no issues and been able to refocus that time on other areas. There is a risk, but no guts no glory.
The last tip is that the safe bet isn't always the best bet. There used to be a saying "You don't get fired for buying IBM". Many CIO's still feel that way, though the company name may have changed. While that may be true, you also don't stand out being part of the crowd. If you want to be exceptional you need to take risks, which may mean not going with the same vendor everyone else goes with.