I have been challenged by a big manufacturing company, which consists of dozens of individual plants (through acquisitions but still function independently with their own data centers and some with their own DW) to build an enterprise data warehouse (EDW). No stakeholders have been identified. Pretty much it's an initiative which has been sponsored by a global business unit. Where should I start and how can I prove the need and importance of methodology. Note: we have been told that no business or functional specs will be provided.


Scott Howard’s Answer: Actually, the fact that you will not be provided business or functional specifications could be a blessing. I sense that your global business sponsor does not want to influence you with a predetermined plan for even the appearance of one. This is the best way to develop a true enterprise integration motivated DW. You have the advantage of discovering the business process models and data models for each of the business units that you’ll need consolidate into the EDW. You then need to reconcile those into a true enterprise standard model, both process and data so your global business unit gets a true conformed view of the entire business. If this is the case, it is fantastic, because that type of a project is what enterprise data warehousing is all about! Yes methodology is crucial. You’ll want to design a multitier warehouse with a consolidated EDW as the central hub eventually feeding all business unit or plant sub warehouses or data marts. This is the best way to ensure consistency.

A word of caution before you begin. A truly integrated EDW will force most of the plants or business units to loose some control over their individual financial reporting standards, and some of their independent business controls. Again, that’s one of your project’s objectives, to standardize these to a common enterprise standard needed by the global business unit. They need to cut through the inconsistent business and reporting standards and accurately compare business units and plants, recognizing unit efficiencies and successes while simultaneously exposing the need for improvement. These business managers will not take this loss of control lightly. You’ll need to come up with an agreed upon enterprise standard for common business measures, common algorithms, and most importantly, common dimensions. You can imagine the chaos should each business unit load their own unique and unreconciled measures into a common EDW. Let’s assume for a moment that the measures are indeed standardized. Would analyzing these standard measures between business units with different business operational and reporting standards make sense? Would comparing weekly totals for two business units, one for which the week begins on Monday the other beginning on Saturday lead to comparable results? That’s exactly why the business units also need to agree on dimension standards like time, geography, customer and product related dimensions to name a few.

As I said, the unit managers will not take their lose of autonomy lightly. They will want to retain their own standards and processes even though the only justification for nonconformity may be "we’ve always done it that way." You’ll need the support of an executive or a well placed project office to help you reconcile the data and processes between each business unit. Someone who can step in when you with all your political clout, or lack of it, can not compel business units and individual managers to conform for the good of the enterprise. Enough on the need for executive sponsorship. Get that stakeholder before you proceed!

Joe Oates’ Answer: If I understand the situation, it seems like a challenge that you should decline. Even though you said a global business unit has sponsored the initiative, it does not sound like this is a true sponsor. A sponsor is a person who is high enough in the organization that can cut through the red tape and politics so that the data warehouse team can do their job. The sponsor also should have something significant to lose if the effort is a failure. I infer from your question that no such person has been identified. One of the most common reasons for failure is that there is no sponsor of the appropriate level and reputation.

Another couple of red flags are that there will be no business or functional specs and upper management has to be convinced of the necessity of a methodology. Unless you can read the collective minds of top management, and anyone who can sign off on the project, it seems as if you and/or the project are being set up to fail. Besides, if you have not tried to build two or three enterprise data warehouses, odds are that the first try won’t be what you hoped it would be.

Clay Rehm’s Answer: Quite the challenge! It actually sounds like fun, but beware – this could be a career ender. This all sounded quite possible until I read the last sentence – no business or functional specs will be provided. Why not? It is in your best interest, and that of the project, to find out the answers to that question.

Don't be afraid to ask questions. Ask why over and over again until you are comfortable with the answer. Document the answers you receive and create a project description with your interpretation of the project based on what you have heard. Work with the global business unit and create a conceptual data model, which is the view of the enterprise.

I strongly recommend you look into the Project Management Institute (www.PMI.org) for help. The Project Management Methodology they provide applies to all projects including data warehouse projects, and the issues you are dealing with.

I also recommend the following books: Radical Project Management by Rob Thomsett – Yourdon Press and Data Warehouse Project Management by Sid Adelman and Larissa Terpeluk Moss – Addison Wesley.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access