“
How can you run Analytics “as one”? If you leave Analytics to IT, you will end up with a first-class race car without a driver: All the technology would be there, but hardly anybody could apply it to real-world questions. Where Analytics is left to Business, however, you’d probably see various functional silos develop, especially in larger organizations. I have never seen a self-organized, cross-functional Analytics approach take shape successfully in such an organization. Instead, you can expect each Analytics silo to develop independently. They will have experts familiar with their business area, which allows for the right questions to be asked. On the other hand, the technical solutions will probably be second class as the functional Analytics department will mostly lack the critical mass to mimic an organization’s entire IT intelligence. Furthermore, a lot of business topics will be addressed several times in parallel, as those Analytics silos may not talk to each other. You see this frequently in organizations that are too big for one central management team. They subdivide management either into functional groups or geographical groups. Federation is generally seen as an organizational necessity. It is well known that it does not make sense to regularly gather dozens of managers around the same table: You’d quickly see a small group discussing topics that are specific to a business function or a country organization, while the rest would get bored. A federated approach in Analytics, however, comes with risks. The list of disadvantages reaches from duplicate work to inconsistent interpretation of data. You can avoid these disadvantages by designing a central Data Analytics entity as part of your Data Office at an early stage, to create a common basis across all of these areas. As you can imagine, such a design requires authority, as it would ask functional silos to give up part of their autonomy. That is why it is worthwhile creating a story around this for your organization’s Management Board. You’d describe the current setup, the behavior it fosters, and the consequences including their financial impact. Then you’d present a governance structure that would address the situation and make the organization “future-proof.” Typical aspects of such a proposal would be The role of IT as the entity with a monopoly for technology and with the obligation to consider the Analytics teams of the business functions as their customers The necessity for common data standards across all of those silos, including their responsibility within the Data Office Central coordination of data knowledge management, including training, sharing of experience, joint cross-silo expert groups, and projects Organization-wide, business-driven priorities in Data Analytics Collaboration bodies to bring all silos together on all management levels
”
”
Martin Treder (The Chief Data Officer Management Handbook: Set Up and Run an Organization’s Data Supply Chain)