Richard Warr (8/20/2013)
I use DQS quite a lot and those steps don't bear much relation to what you do.
1) Create/Map Knowledge Base
2) Create DQ Project
3) Run project to cleanse data
is more like it.
There is a "Configuration" option but it's not always necessary.
Nice to see another of the features introduced in SQLS 2012 making an appearance in QOTD.
I suspect that the question's first two steps are your first step, while the question's third step is your second and third; but I'm not at all sure, as I've done very little reading on the topic and I've never used this feature. I got it right because I had a vague recollection (quite probably incorrect) that there was something about first configuring the source of valid data, then creating a domain and populating it with that source, and then using the result to do the cleansing. The documentation for DQS at the conceptual level is (as it so often is for MS stuff) pretty abysmal, but probably the lower level documentation is rather better (or nobody would be able to use the feature). If the source is a provider that you subscribe to and have configured before, configuration is irrelevant because it's already been done, so the questions first step doesn't happen, and you go straight to creating a domain and mapping the data source to it; if you are providing the reference data by hand, there isn't any configuration to do as far as I understand it. So the apparent difference between the qotd and your experience is the real difference between two different sets of circumstances in which someone uses DQS.
At least that's how I understand it - I could easily be completely wrong.
edit: I've now seen Hugo's comment. The awfulness he describes may be one reason why the concept-level documentation is so utterly awful. I suppose I'll suffer it one day, but I shall be optimistic and assume I don't need to play with it until it's proven that I do.