Five Things You Need to Know About Your Users BEFORE You Deploy BI – Part 2 – The Third question.
This is the third of five questions you should answer before beginning a BI implementation. This one will require a little extra work as it has multiple layers.
What types of questions will users be asking? (This differs greatly from “What information do they need?”)
Have you ever really thought about the questions that people ask? Can they be categorized? If so, how?
Different types of questions are best asked and answered by different types of tools. Performance management, dashboards, and scorecards are about monitoring the status of a key metric. While ad hoc query tools are ideal for asking random or on-the-fly questions.
This list describes the various kinds of tools available and highlights the types of questions that each is best for answering:
Status is measured regularly. Users know this measure exists and where to get the data related to it. Therefore, you want to make it easy for them to access through pre-built reports or dashboards.
A variation of status where similar or related data is required, but is not readily available through a pre-defined dashboard or report. This is known as an “ad hoc question.” It is investigative, meaning that some knowledge or curiosity has sparked the user’s interest. He or she knows that the answer exists, but does not have a way to retrieve it automatically. This type of question is one of the most challenging, because it will have multiple answers, but each answer may or may not be practical, depending on the sophistication of the user asking the question.
A business intelligence tool like ad hoc query can certainly solve the problem, but not all users have the skill or the time needed to use such a solution. An alternative approach is called “guided ad hoc,” which is comprised of one or more parameterized report templates. These templates allow a user to customize a pre-existing report as a means of answering an ad hoc question.
Sometimes, users are unsure or unaware of the existence of something (e.g. a product or a customer), and will need to explore or discover data to find the right information. The question they are asking is, “Does XXX exist, and where I can find it?” which is a much more common question than you think. It is asked countless times every day through search engines such as Google. Even in corporate scenarios, this question is best answered with a search-style interface. Users type in a word or phrase to see if there are any matches within the databases or applications that reside across a company.
Show Me What’s Interesting
This is one of the more unusual questions, and is typically posed by the curious people described earlier. Often it is the responsibility of an analyst to find small issues before they become big problems. Users will say, “Look at this data set and tell me what I should be concerned about.” They want the analyst to point out what is unusual, so they can begin to ask more educated ad hoc questions. These inquiries are best answered by visualization software that presents large or complex sets of information in a more intuitive, graphical way.
Predict (What Should I Expect?)
In the past, predictive analytics were reserved for statisticians. But this analysis method is becoming more popular as business intelligence vendors find better ways to embed it in operational applications. When the two technologies are effectively combined, predictions can be applied to nearly all other forms of information delivery. They can be incorporated into dashboards or reports, or customized via guided ad hoc interfaces.
While the creation of a predictive model still requires the assistance of a specialist, the results of those efforts can become part of any frontline processes.
Stay tuned for part three of this series - The final two questions.