In our previous post on pre-built analytic applications, we described a framework for choosing between custom and pre-built business intelligence. This post describes common pitfalls in implementing pre-built applications.
Pre-built analytic applications are sold by their vendors as a BI strategy that is very quick and easy to implement. This often causes user organizations to underestimate the effort it will take to drive business value. Although these applications can be a very valuable jump start to a business intelligence effort, implementing them is still a large, complex project. Here are five key things that often go wrong:
1. Believing you do not need a BI/DW team
This is the biggest mistake organizations make. BI/DW teams are expensive, and avoiding that skilled labor is a big selling point of pre-built tools. However, without a skilled team, you can not fix the application when it breaks or customize the application to reflect changes in your business. Without skilled business intelligence professionals, there is no basis for checking the vendor’s claims about system functionality or implementation times.
Making it Right: Hire at least one business intelligence specialist with prior experience in the tools you are working with.
2. Planning to ‘finish” the application
Even pre-built applications are never truly complete. Reporting needs change, the organization changes, and the underlying source systems change.
Making it right: Budget for an initial big implementation, but also budget at about 1/3 that level for ongoing changes and improvements.
3. Not testing
Pre-built applications contain bugs. They also may reveal problems in your main business systems that contain the source data.
Making it Right: Plan to validate all the data in the new system. Budget time for data fixes and fixes to your operational systems.
4. Overly-aggressive time frames
When a vendor touts a six week implementation cycle during the sales process, this often means six months in practice. In particular, even if you can roll out the technology that fast, users may not have the capacity to test, train, configure, and adopt. Tasks like procuring hardware and scheduling end-user training take calendar time rather than working time.
Making it right: Plan to overlap several smaller projects so you have something to work on during delays caused by external factors.
5. Not using outside help
Two forms of outside help are important. 1. Support from the vendor. Vendor support has unique product-specific resources, but tends to be expensive and not customized to your business. 2. An implementation partner. Vendor support is about selling licenses. A third party gives you a broader perspective, an honest broker, and access to industry and process skills.
Making it right: Include vendor support in the big initial purchase, when you have the best bargaining position. Sales engineers are the stars of the vendor’s support organizations and therefore hard to get. If you work with a particularly good one, name him or her in the contract. We recommend selecting an implementation partner before the initial implementation, and continuing on a limited basis during follow on work.
Pre-built analytic applications can jump start your BI efforts. With planning and the right resources, you can realize significant savings and faster implementations.
by Tom Victory, Principal Consultant