1) Politics
a. Reasons
i. Turf battles in the team/between teams ii. Vision and focus of the teams iii. Certain people may feel threatened by the outcome of the project iv. Claims of owning the data and reluctance in sharing it
b. Solution
i. Strong willed project manager ii. Have clear vision of the project outcome iii. Have a business champion cut through the red tape and push the project forward
2) Inadequate user involvement
a. IT department tries to work on the project in vacuum(aloof from business)
b. Not focusing on the business requirements phase, leading to complex and hard to use business systems c. Solution
i. Involve users early ii. Work with users to help them understand how to formulate questions iii. Show users frequent updates to make sure you are on path to solving a problem iv. Establish a training plan for end users
3) Trying to build the “Mother of all Data warehouses”
a.
b.
c.
d.
Trying to build an enterprise data warehouse (boil the ocean approach)
2 schools of thought – top-down and bottom-up. Top-down is difficult to implement
Takes long time to deliver and cost is high
Difficult to use lessons learnt, because you cannot learn as you go along in a phase( as opposed to building phased rollouts where you learn lessons in each phase quickly)
e. Solution
i. Start by building data marts. Methodology is same, but scope is different. ii. Gives quick hit wins and helps secure funding for future releases iii. Initial projects do not take more than 3 months (generally)
4) Incorrect or inadequate KPIs
a. Even simple cubes do not give valuable information, if not used by business users
b. If built by IT only, then there are chances of delivering partial or incorrect information
c. Solution
i. Work closely with users and business champions ii. Consider delivering data, based on the users’ requirements(either reports/scorecards or analysis) 5) Poor data quality
a. Building a data warehouse is the best way to discover bad data and operational problems. Cite examples of sales people hierarchy, where sales guys reported into divisional managers bypassing the regional managers.
b. Identify a resource who can own the data and verify it’s accuracy
c. Create unknown buckets in the warehouse, so that such bad data can get visibility
d. Use data warehousing projects to drive fixes in your operational systems (depending on the tolerance levels for bad data)
PDF created with pdfFactory trial version www.pdffactory.com
6) Overtly complex cubes
a. One solution for all the business problems
b. High number of dimensions tend to discourage users
c. Solution
i. Remember that there are 2 types of cubes
1. Reporting – used by senior management in the form of KPIs
2. Analysis – used by data analysts ii. Build simple cubes or simplified views of these cubes, based on users
7) Inadequate funding
a.
b.
c.
d.
Funding is an issue, if estimated