There's some interesting studies on this I read somewhere.
I think the biggest thing is to learn as much as you possibly can before you do anything significant. The biggest temptation for leaders is to come in and be the leader by stressing big changes. This causes a backlash against the leader. The very best thing you can do (assuming you've already got/accepted the job) is to learn.
Set up meetings with every single stakeholder in the organization and ask them what they would love to see from the community. Setup meetings with your best members and a few regular members. Find out what they would most love to see.
The most people feel you listened to them the more they will listen to you later. Don't try to make any big changes until you've done this bit.
The next stage would be to benchmark where you're at now. Get all the data you can and benchmark your community to check the ROI, the level of growth, activity, etc...benchmark member sentiment too. This will be really useful to prove your value later on. Think this echoes what @dun3buggi3 was saying too.
Only once you have done this would I begin making tweaks/changes in what you're doing. I guess the saying is true, first seek to understand and then be understood