subscribe: Daily Newsletter
search the site
Why BI projects don’t and won’t live up to expectations
A recent National Computing Centre (NCC) survey created something of a stir in the UK when it found that 87% of business intelligence (BI) projects in that country fail to live up to expectations, writes Richard Greyling, managing consultant at Knowledge Integration Dynamics (KID).
But I'm not surprised by the findings.
The NCC claims to be the biggest IT corporate membership body in the UK championing effective IT deployment.
Its survey findings include some other interesting results:
* Only 13% of respondents feel they have achieved their BI objectives completely;
* 22% consider their company's BI performance to be lower than expected; and
* 24% reported that IT development and support costs have exceeded expectations.
It suggests that the primary method for reversing the situation is to resolve cultural issues.
Still the biggest single reason for embarking on BI projects is improving decision-making quality. Improved performance measures, data accuracy and data integrity rank second and third, while being more responsive and flexible for introducing new products and services to market didn't really feature.
So where have BI projects gone wrong?
The oft-touted reasons are:
* Organisations don't incorporate the three critical success factors into their BI projects: aligning data with business requirements; choosing the right tools and technologies; and aligning people, process and culture;
* BI projects get hung up on technical aspects too early on with companies trying to ride the technology wave;
* Figuring out an appropriate technology fit; and
* Inability to access the correct detail frustrates users and hinders adoption
But is that true in South Africa?
The only reason BI doesn't deliver is misinterpretation of what the solution will do. People expect that it will revolutionise their business and do everything for them in one fell swoop. Business people want to take a big bang approach and get all the functionality immediately. But the only way to successfully deliver a BI project is iteratively.
I know of large local corporations that have very successfully delivered BI projects. In one case they took all the appropriate steps. They appointed a senior person to take care of their side of the project. They took an iterative approach that saw a basic BI sales solution rolled out across multiple divisions of a very large operation followed up with an optimisation phase. That was successful because they included basic, common functionality for all the divisions in the initial phase. The optimisation
phase then delivered specific functionality for the salespeople in the different divisions.
Companies that don't take an iterative approach like that are going to find themselves incorporating too much functionality in the first build with a resultant increase in complexity and potential for error. It's just too complex and unwieldy and bound to fail. And technical people rarely counsel that it cannot be achieved because to them it's a technical challenge that they want to meet.
Another factor South African business must take into account is that, considering the rand/dollar exchange rate, BI projects are not cheap. One of the ways people try to save money is by having consultants investigate their needs and design a solution which they themselves try to implement. Nine times out of 10 that's not going to work because they simply don't have the in-house expertise. I've seen many cases of companies trying to do just that and then having to call in the experts afterwards to rectify the situation.
Another attempt to work around inadequate budgets sees businesses implementing only portions of BI suites or solutions recommended by consultants. But the result is that they end up with an ineffective tool not suited to their requirements.
Executive buy-in is often quoted as a reason for BI project failure. The reason for that is not necessarily that the management team doesn't see the need for the BI project and push that view down through the organisation. Often it means that the IT department was left to implement the solution, which means that it comes entirely from a technology point of view. The IT department implements technically efficient tools that users don't want or that don't give users the view of their business that they need. BI is a business solution that must meet business needs and must be sold to the business before tools are chosen, technical challenges dealt with and the implementation project kicks off.
A critical failing of the BI industry in South Africa currently and in the past has been dropping boxes, freeware, bundles and the resultant shelfware.
In the past resellers dropped boxes. Clients ordered BI software and a sales consultant dropped the package off with the IT department. There it often sat, unused. Technical staff either wouldn't know what to do with it or the business didn't want it. The result was that the package was dropped on the shelf and became shelfware.
Free reporting software was liberally distributed in South Africa in the mid- to late-'90s and dubbed BI. But they're essentially two different things. Reporting is simply a list of information used for operational purposes. Reporting tools in the past tended to be slow and didn't deliver much, if any, BI functionality. Today they're called dashboards and they sell well at the CXO level but as soon as the decision to buy is made, then the vendor must deliver real BI capability at lower levels of the customer's organisation. That means anything up to online analytical processing (OLAP) cubes, which is a complex environment, and can lead to CXOs becoming overloaded with information.
Bundles are another problem today. Some database and ERP vendors bundle what they call BI solutions with their core software. That again either leads to shelfware or failure to meet expectations. BI is very different from ERP and databases and should be handled accordingly.
In most cases the local industry has matured beyond the box-dropping mentality but many problems, such as the bundles and freeware, remain. Until the industry matures to the point where BI is accorded the individual treatment it deserves and customers are educated about the pitfalls of their decisions, the current situation won't change.