If you haven't looked at how your IT division is handling master data for business intelligence (BI), you should probably check under the hood: This seems to be a problem area for many organizations, and as with all-things-data, the problems eventually erupt into a big, bad tangled mess.
Gartner's Andrew White wrote about the "fudging" of master data management (MDM) that tends to happen with BI, aka analytical, systems. But from his post, I gathered this was more of an occasional problem-a sort of fable about what not to do.
However, it's very clear after reading this Information Management article, "How MDM Changes BI Best Practices," that this problem is pervasive, emerging from a common BI practice of dealing with master data inconsistencies downstream in the extract, transform and load process.
The writer, Michael Vlund, is an IT services consultant specializing in BI. He suggests it's time for BI to embrace real MDM. The first half of the article makes the case for why this is an important issue for companies and BI practitioners to address. The second part takes a very high-level look at what you'll need to consider as you address it, as well as the benefits of moving MDM out of ETL:
"There are great benefits to be harvested by the BI departments that embrace the MDM discipline and that revisit the architecture and ETL processes from an MDM perspective. Transferring mitigations of relevant master data-related issues from the ETL jobs to the transactional MDM capabilities can lead to greater flexibility, improved performance, faster implementation cycles increased transparency and maturing the organization in understanding the link between data in the operational environment and the reports they are getting."
To be honest, it's not an easy read, but it seems like required reading for those concerned with BI or MDM.