Detect Anomalies Like Exceptional Swings Accross Aggregations?

Hello,
Not sure if data minimg can help me solving this type of inquiries. Thanks in advance for any pointers you could provide me.
Problem:
We have many dimensions with many levels. Sales directors will sometimes note a drop or a spike when looking at one of the dimensions members, most of the time at a high level.
Example, For Q4 07, the aggregation of Q4 Billing plus remaining Q4 Backlog dropped by $10M in America for the Automotive business from week 4 to week 5 while it was forecasted to remain stable.
The immediate question that will be sent to the guys in charge of this business is "What happened?".
Obviously evyone struggle to answer these questions eveytine there is a drop anywhere.
Using classic tools, they will try to spot a single anomaly or determine if this is accross the board and what are the concurent factors like lower than expected shipments combined with some orders cancellations.
The drop-off can lie in a single product or customer location, in which case it is fairly easy while time consuming to spot the anomaly using classic drill-down approach.
The drop-off can also be more wide spread or caused by a combination of factors like unable to ship in time, orders cancellations and sales credits in which case it is more difficult to spot where is the anomaly and explain it.
So I wonder if there is any way that I could use data mining to make the entire process of detecting anomalies accross dimensions levels, members and measures more straight-forward for end-users.
What do you think?
Thanks,
Philippe