The Big Picture: A BI Project is Like Building a House

When you build a house, you start from the foundation and work up.  When you build a BI solution, it is logical to start from the foundation and build up as well.  However, what I see often is someone working on the house roof (reporting) before there is a foundation (data model, integration, security ...).

We all understand that some houses are pre-fabricated and the individual pieces are build independent of each other and then somehow come together in one miraculous final push to production.  However, in my opinion, that is not the "industry standard" and as a BI consultant, I will rarely recommend any other BI build method other than "from the ground up". 

I consider the house roof, reporting and analytics, to be the "fun" part of every BI project because it is the most visible.  (I have observed that those who get to write reports and create dashboards often progress quickly to hero status.) However, it is the responsibility of each team member of a BI project to produce a product that has these characteristics:
    1.  Accurate (is dependable and truthful)
    2.  Scalable (can grow and change in step with business fluctuations)
    3.  Discoverable (you can find things you want, such as metadata, measures and attributes)

Talking Points:

1. If you have data transformations happening in SSIS, your EDW views, your SSAS MultiD and tabular model DSVs, in MDX and DAX formulas,, and in Excel Power Query and PowerPivot, how accurate do you anticipate your DW information to truly be?  Where there is a question and you have to prove the actual version of the truth, how many places do you want to 1.) search and then 2.) fix?
   
2. If you have star schema, snowflake schema and 3NF (3rd normal form) data models in addition to aggregated data marts and disparate data stored only in Excel or SharePoint lists (for possibly very good reasons), how easy will it be to add a new subject area, corporate department, or newly acquired company?   
   
3. Very few companies can build the Taj MaBIsolution right out of the gate.  Often MDM (master data management) falls prey to budget and time constraints.  However, we are building a house here and if my house has four kitchens (customer tables), where should we tell the marketing department to eat their lunch (discover the most recent customer data)?
4. In MS BI, there are many ways to get to the same finish line, but where will you decide is the proper place for ...
   

                    Data integration -- SSIS, SSAS, views, USPs, UDFs, DSVs, Excel
                    Self-Service BI -- PPS (performance point services), Excel, Report Builder
                    Automated Reporting -- SSRS, Excel
                    Source code control -- TFS (team foundation server), corporate file store
                    Documentation -- in-line code, TFS, file store, DMVs, Visio, Word documents

---