Outsourcing Your Data Warehouse

[Steve Bennett]

I was intrigued by a recent post on the CORTEX about Optus signing a three year deal with HP Enterprise Services (formerly EDS) to outsource several core applications including the data warehouse. I don't have much more detail than that but my first thought was to wonder about the wisdom of an organisation outsourcing core capabilities like analytics. Later on I thought a little more about this and I wondered if my first reaction was wrong - and hence this post. I believe that an ability to compete on information (i.e. analytics) is a core capability that many firms have to possess to succeed. Does that mean that they also have to keep in-house all their information management activities? Is the entire information chain from source to decision a core activity? Or can you outsource the data warehouse/data stores and still be good at 'doing analytics'? I don't have any hard evidence either way so I will make a judgement call and say no. What's my justification for this? Here's my logic: A warehouse is a lot more than a place to store data. Building and operating one means that you learn a lot about your data, how it is used operationally, what it is - and frequently what it is not. I'm talking here about the quality of your data and the insight that this knowledge gives you about the soundness of the data you use to make decisions. The people that make ETL happen can also tell you a lot about what really goes on in your transactional applications and business processes. The best ETL and data warehouse experts I have worked with have always had a fantastic ability to support business people and ensure that they become better information users. Sure, you still gain insight from your analytic activities that make use of the data warehouse but I would worry about the impact of outsourcing on my business decision makers ability to access the knowledge locked in the (HP) ETL and data warehousing team. This may not be as important in organisations with well developed MDM infrastructure that robustly captures and makes available this information to the business. I'm pretty sure that few such organisations exist today. So Optus will (presumably) save some dollars in the short term but as the months go by the hidden costs associated with less insight will greatly outnumber these. Ultimately, this may all be a moot point as based on discussions I have had with Optus BI staff over the years, as an organisation, they do not seem to be very good at analytics anyway. A lot of good analytics people have left Optus over the years and they all seemed to be frustrated at the inability to make an impact on this. I'm also aware of the irony of my opinion being based on 'experience' when I spend my career advocating fact-based decision making. So let's just keep this post between you and me. OK?



Get More from the original blog...