The intellectual aspect of business intelligence (BI) can see the importance of proper data models, best-practice ETL processes and optimised cube or reporting structures being overstated.
The BI environment is simply a means of migrating data from A to B to improve the quality of analysis and reporting, or, in other words, to work towards that ultimate goal: actionable intelligence.
BI in itself is a taxing topic; it resides on the intersection of business strategy, business processes, requirements, and available IT assets (internal and external).
The parties involved include business and IT sponsors, end-user communities, project teams and IT professionals. If you lose yourself in the process and consider the means a goal in itself, projects can take too long and cost too much without delivering the expected merits to end-users, resulting in long faces.
By overstating the data warehouse side and the technical BI aspects, there are no clearly beneficial results because it is too difficult and end-users are ultimately unhappy. In essence, there's a clear gap between what users want and the solutions that are delivered.
How can we make BI fun, fast and easy to use? OLAP Report author Nigel Pendse mentioned the "Google effect" in his presentations during his recent South African tour. Performance is of the utmost importance to end-users. How can you expect end-users to have fun with BI when they only see results in 10 seconds, whereas Google shows you results in a fraction of a second? BI needs to stay in touch with this reference if it's to be fun for users.
And that while querying a database that is an awful lot bigger than the average BI database!
But it's important to bear in mind that Google has a lot of hardware to support its response times. The business invests vast sums in its hardware infrastructure, something most organisations cannot afford. However, even though hardware is a factor in performance, the non-hardware-related factors are quite often more important when it comes to performance.
Performance can be a difficult topic to grasp, but there are a couple of tips that have proven very useful:
* When using OLAP technology (cubes), make sure that as many aggregations as feasible have been designed, using pure MOLAP technology (multi-dimensional OLAP);
* Aim for small cubes. Don't try to put too much in one cube; you want small cubes even though the technology seems capable of far larger cubes. Performance will be negatively influenced if there are too many dimensions and measures in the cube, as well as too low a level of detail. Cubes are ideal for viewing aggregated data;
* As soon as detailed data is required, use a query and reporting tool like Microsoft Reporting Services to show a standard report with the details. It works like a charm if you automatically pass on parameters and filter settings from the cube to the report, so the integration between the two appears seamless; and
* From the start of the project, listen to what the end-users really want
Key for any BI environment is end-user adoption. It's mission-critical to any BI project. If end-users don't work with the system, then it's a futile and wasted effort. If end-users want the solution, then it will be much easier to implement, it'll be more fun for them to use because they have provided a lot of input into the environment, and they feel they own the solution. This in opposition to IT departments creating reports and cubes and not the end-user community. User involvement in the design and development phases is crucial to adoption and champions are critical.
Change management will help with end-user adoption. In this regard, it is more fun dealing with a solution which you understand and which really leads to results. All too often, end-users are afraid to say they don't understand the solution (but who will be brave enough to say the emperor is not wearing smart new clothes?).
Short, easily defined projects
Finally, the chances of success will be enhanced by focusing on short, easily defined projects that have a high impact on the business, and delivering the right end-user deliverables.
BI's challenge is that it's not seen to be required like transaction environments are because it generally does not support primary processes. The primary business operations, such as orders, invoices and stock are supported in ERP systems and BI sits on top of that. If operations are running smoothly before BI is employed, then users question its necessity. That's the BI challenge. Ensuring that users see its necessity to the business and to their roles in the organisation.
Organisations often don't spend enough time marketing the solution to users. There is often a sense of urgency to implement BI solutions, but when end-users don't perceive the need, and when organisations don't slow the project down enough to ensure successful marketing to end users, then they're doomed to failure or partial success. BI makes users' lives easier. They get the tools to use the data content that positively impacts their performance. And that's why organisations should market the successes they've experienced, focusing on how individuals have benefited.
It's called change management and it's an integral component of any successful BI project. Companies can actively ask their user communities for better ways of dealing with the data, solving specific problems and sharing what works for others. Too many programmes focus on the technicalities instead of how BI can be fun, fast and easy to use. Keeping it simple is the best strategy. It puts users in the driver's seat and when they're in control they start to explore BI, uncovering the benefits for them, they take ownership by determining the content and with that awareness, the fun factor emerges too.
Big-bang projects cannot employ this approach. But iterative delivery against a proper business case with clear focus will result in goal-oriented achievements.
Share