Like some others posted, the 7-10 range is to a large extent subjective.
Anything below 7 should start to trigger red flags.
Overall, assuming you haven't found a single source with whom you tend to really agree, it's a more stable indicator of the average consensus as it's an aggregate result that takes several sources.
Another cool things it does is that it gives you an aggregate result for a whole bunch of critics and an aggregate result for a bunch of users which allows you to discover bias critics might have.
People like to bash on game critics, but most of the time, the aggregate critics score and the aggregate user score are pretty close which indicates that what critics say is a decent predictor of the average among non-critics.
I usually check the aggregate score on metacritics (both for critics and for users) before buying a game here, mostly to filter out those that are rated less than 7.
EDIT:
As I pondered about it, something else occured to me: Not all bias in a rating system is bad.
For example, a strategy games will be mostly reviewed by users who have at least a casual interest in strategy games.
Assuming you know yourself enough to know that you enjoy such games, you probably don't want low review scores from people who don't like strategy games polluting the average, because what you're really interested in is knowing what people who have at least some interest in strategy games are saying.
Post edited July 18, 2015 by Magnitus