In-Depth
Q&A: The Power of Dynamic BI
Dynamic business intelligence can quickly deliver answers to business questions, but it can be difficult to implement in today’s technology and business environment.
Dynamic BI describes information that is ready when the business needs it. That might be in a month, but increasingly that means days or hours (or even sooner). Cliff Longman, CTO of Kalido, says that although dynamic BI offers many benefits, implementation is challenging. Longman says that ill gradually change, leading to information that is truly dynamic -- accessible to business leaders precisely when they need it.
We spoke with Longman about the nature of dynamic BI, what prevents adoption, and how business modeling tools can help an enterprise adopt the technology.
BI This Week: What is dynamic BI?
Cliff Longman: The short answer is this: Dynamic BI is business intelligence that is adaptable enough that it can deliver answers to business questions in time for those answers to be useful.
How dynamic your BI needs to be depends on the business questions asked of it. Some questions might need answers within an hour, others within a month. Your current BI infrastructure may be able to answer some questions while others will push the infrastructure to a breaking point.
One thing that's interesting about traditional BI is that in many cases the business questions that BI tools answer don't change much over time, if at all. What changes is the information that feeds those questions. "What's my gross margin?" might be a question you ask every day; the answer changes dramatically, of course, as you acquire a new product line, reorganize your divisions, or change your pricing model. Dynamic BI has to be able to cope with changing data that reflects the everyday changing business world.
In addition, questions involving a specific project, a new business opportunity, or a significant change in the business environment can often arise at a moment's notice and change just as quickly.
Dynamic BI goes beyond traditional BI, delivering answers not only to the questions the business asked when the BI system was first installed, but also to new questions that come up daily: How does our company look post-merger? What would happen if we changed our sales territories? Is my current marketing promotion having an impact on sales?
How difficult is dynamic BI to implement?
Many or most traditional BI implementations have difficulty delivering dynamic BI, since every change requires some reconfiguration of the data structure behind the BI tool. Often, by the time the system is ready to provide answers, it's irrelevant -- the opportunity has passed or decisions have already been made.
For example, say you are a toy manufacturer. Thanks to a product placement on a popular TV show, demand skyrockets for one of your action figures. Now you have to answer a whole new set of business questions: Do you have enough manufacturing capacity to support an increase in production of this action figure? Should you transfer production capacity from other action figures to the hot one, even if demand for it drops just as suddenly as it has risen? Could this hot product provide the company with an opportunity to expand its market share in the action figure market?
Without dynamic BI, getting those answers could take weeks or months, by which time you've already moved on. Dynamic BI can provide you answers in days or even hours, keeping up with the speed of your business.
What is preventing companies from securing dynamic BI?
There are two basic reasons: First, despite exercises in requirements-gathering, it's often difficult for IT and business managers to ensure that the BI infrastructure truly reflects the business model. That's partly because business managers can't easily understand a data model, and therefore can't validate it or suggest changes. As a result, data warehouses are often built in ways that do not reflect the business.
The second reason is that, even when the data structure reflects the business, the data warehouse infrastructure supporting this structure is rigid and difficult to change. Therefore, when the business environment changes significantly, and a new BI data warehouse structure is needed to support the new questions raised by these changes, the old data structure has to be manually changed, reworked, and potentially rebuilt, which takes time and money.
Ironically, it is during these times of change that business intelligence can prove most valuable to business managers -- if the intelligence arrives in time.
Why is it so difficult to restructure the data warehouse for new BI questions?
While it might seem simple, making even small schema changes to a database structure's basic design can have a significant impact on all the other parts of the data warehouse system, from loading to validation, exception handling to storage, summarization to BI queries.
Moreover, IT often has a problem understanding the changes that the business managers need, and the business managers often have difficulty describing these changes to the IT staff. So you often end up with the IT staff restructuring the data warehouse, only to find that it is still unable to provide business managers with the BI answers they require.
Why are audit trails that track changes to a data warehouse's structure so important when implementing dynamic BI?
There are two main reasons. One is regulatory compliance. If you're asked for the origination of a number or other data used for financial reporting (even if this was not its original intended use), compliance regulators will almost certainly require that you can trace where that number came from.
The second is time variance. If you've made an acquisition in the past year, the finance department may well want to see revenue breakouts both pre-merger and post-merger, comparing like for like, or your sales department might want to see revenues of a particular product line based on how the company was structured at the time of the transaction (even though that product has since been moved to a different brand due to the acquisition). Meanwhile, executive management may want to see what would happen if they reorganized the merged company's product lines even further. If you don't keep track of all data changes, you won't be able to provide all these departments with the answers they need for the time frame they need.
How can business modeling tools help companies achieve more dynamic BI?
Business modeling tools capture the way a business works in a medium that can easily be understood by both business managers and IT. With translation barriers eliminated, BI projects are more likely to address the needs of business managers out of the gate, and can easily be modified to address future changes.
More advanced business modeling tools automate dynamic modeling of a data warehouse's data structure (Kalido's Business Information Modeler is one), making the translation from business model to data model to data warehouse structure nearly automatic.
Perhaps the most important point is this: If the data structure is actually based upon and driven by the business model, then changes to the business model can be automatically reflected in the data structure. This limits the issues that come with changing a data warehouse's schema, and speeds the delivery of BI system answers to new business questions. This approach makes business intelligence far more dynamic.