Internet search has dramatically changed how the world accesses information. As a result of popular Internet search engines such as Google, nontechnical users have become accustomed to getting instant access to information. In the workplace, nontechnical users haven’t faired as well. Many business users still struggle with how to get the answers they need when they need them. The challenge is that the information they need isn’t just an Internet search or mouse click away. Business answers gleaned from knowledge data sources by sophisticated business intelligence (BI) tools are often only accessible by a small group of power users, such as the corporate IT staff or analysts.

 

Many organizations are struggling to address this important issue. As the quantity and complexity of business information grows within the enterprise, it is increasingly important to provide employees throughout the enterprise with quick and easy access to the right information. In today’s highly competitive corporate environment, self-service BI is no longer viewed as a nice-to-have, but as a requirement.

 

Imagine a world where employees could access and analyze corporate business data, reports and metrics just as easily as they search the Web. Ask a question, get the answer. By applying the metaphor of Internet search to BI, that world is within our reach.

 

Dissecting the Commonly-Used BI Attribute: “Easy to Use”

 

To meet the goals of pervasive BI, a BI application must satisfy the requirements of a large user base of people with varying degrees of corporate knowledge and computer skills. Add to this the range of vocabulary-related challenges due to the diversity of the English language, and you have some major hurdles to overcome. The application must be easy to use for all people and also must deliver secure, fast, accurate results. This represents quite a challenge in how BI is delivered. Such a tool encompasses the following characteristics:

 

  • Familiar user interface. The BI application’s user interface must not be intimidating. Further, users should know the basic interface interaction methods even before they use the tool. A person can drive any model automobile. A low barrier to use is one of the reasons automobile use is pervasive in society with basic familiarity and no special training.
  • Familiar language. To interact with a BI tool, users should not be required to learn intricate report or query builders or complex query languages such as SQL. Users should interact with the application using their own knowledge and vocabulary.
  • Reduced training. Employee turnover occurs regularly in every corporate environment. Training employees to become competent, let alone expert users of sophisticated tools is costly and a drain on existing resources. Any such requirement presents a hurdle to adoption, reducing the pervasive use of the BI application. An effective BI application should require little training.
  • User assistance. A common refrain from BI search users is “What do I ask?” Staring at a blank search input box can give a novice user stage fright. Like typical Internet search engines, BI applications should provide unobtrusive assistance to users in the form of sample questions, query suggestions and term suggestions to quickly guide the user on a useful query.
  • One-stop shopping. With myriad BI tools available, users often have to switch between applications to uncover answers, prompting the frequent question “Where do I look?” A pervasive BI tool must be the sole application that users interact with for answers. The application must compute answers by drawing upon existing reports, applications, legacy assets and corporate databases.
  • Secure, fast, reliable. Finally, and most importantly, any pervasive BI tool must provide secure, fast and accurate answers every time. Because a pervasive tool interacts with users of all levels, the tool must enforce corporate-mandated security constraints, allowing users to access to only the information they are allowed to see.

BI applications that extend the Internet search paradigm with advanced natural language query technology coupled with query suggestions can meet the pervasive BI challenge.

 

Familiar User Interface

 

The Internet search metaphor can be extended to enterprise business data. Further, by applying expert linguistic understanding to the search query, we can directly address the chief problem: How can corporate users easily find business answers?

 

With a simple, familiar, yet powerful search box, users can easily do ad hoc queries and report searches without cumbersome query builders or the need to understand the underlying data. The result is that people at all levels of the organization can quickly retrieve and share answers to their business problems.

 

For example, users might ask, “Show sales of each category in the last month.” Software linguistically processes this query, making sure to extract the semantic meaning versus simply matching words. After processing the query, the software retrieves the appropriate data, computes the answer, and displays the results in an appealing format - an automatically generated report, a graph, PDF or simple HTML page. In addition, the software searches existing corporate assets such as existing reports and presents what it finds alongside any computed ad hoc query results.

 

Familiar Language

 

The terminology spoken by business users is not the same as that spoken by a CFO or by operational analysts. Each employee may have a unique way of requesting information. By using the power of natural language understanding, the experiences across all types of users are normalized. Identical questions asked in different ways, using different vocabularies, will return the same result. The search vocabulary of the BI application should grow and adapt to the needs of your user community.

 

Below is a comparison of questions entered by power users and business users.

 

Power-User Question: What’s the total value of all deals this quarter with at least 50% probability?

Business-User Question: Deals this quarter over 50 percent.

 

Power-User Question: Who bought personal auto policies with accident benefits but not collision, by branch?

Business-User Question: No collision benefits.

 

Power-User Question: What customer had the most dollars returned in May 2004?

Business-User Question: Customer returns may 2004.

 

Power-User Question: Show the hours, earnings and deductions for Bill Adler’s last paycheck.

Business-User Question: Paycheck info for Adler.

 

Predefined vocabularies can make it even easier for users to rely on a natural language ad hoc query approach. Specialized domain-specific vocabularies for business functions such as sales, finance and human resource BI applications include predefined industry terms or terms associated with a specific role or company. These predefined vocabularies greatly reduce the time to deploy a custom pervasive BI solution.

 

Reduced Training

 

Unisphere Research cites lack of training as the biggest obstacle to making better use of BI, followed by limited staffing resources.1 There may be a dozen or more different BI applications within a corporation, each with a different, often complex, user interface and each with a different training requirement. These environments further complicate the “Where do I look?” and “How do I ask?” questions.

 

How many of us have been trained on the use of Google? Relatively few, and if some were, it likely did not take more than a few minutes. Ease of learning is a requirement for widespread acceptance. Traditional BI tools are complex and require significant training. What if employees could ask questions in their own language, similar to Google, and get business answers back?

 

Modeling BI application interaction as simple to learn and to use as the ubiquitous Internet search engines greatly reduces the training and ongoing maintenance needs of pervasive BI deployment.

 

User Assistance

 

People who deploy pervasive BI-for-the-masses solutions are faced with the challenge of training hundreds, perhaps thousands, of users. As BI spreads through an organization, it is impractical to continually train the entire enterprise on how to interact with a new tool. Users accept new tools more readily if the tools themselves are easy to use and automatically helpful. Through query suggestions, users can learn on their own how to maximize their BI query experience.

 

Query suggestions should start when a user begins typing a query. Using advanced linguistic processing, sample query suggestions and word completions can be suggested automatically through familiar web-browser drop-down menus. As the user enters more text, the software can generate suggestions based on past queries, By using advanced linguistic techniques rather than simple word matching, the software can match user input to previously asked questions more precisely.

 

One-Stop Shopping

 

Applying the Internet search metaphor to enterprise BI offers a unique set of challenges, different from that of Internet search. Specifically, answers for corporate users often reside in structured data sources as well as in existing documents and reports. When using text search technology similar to Google’s to mine a corporate database, two- to three-word searches will return relevant documents rather than the desired specific answers. Often, answers involve more than a simple document found as a result of a keyword search query. Here, deep linguistic processing excels over simple text search. Answers must be uncovered and computed from existing data, possibly from multiple disparate data sources. Answers may be in the form of an instantly computed spreadsheet, chart or graph, or they may be in the form of an existing report created by another system altogether.

 

Secure, Fast and Reliable

 

With every level of an organization able to access corporate intelligence data, security becomes a paramount concern. Pervasive BI applications must provide role-based authenticated interaction in order to allow all users seamless, secure access to critical business data. In short, users must not see data that they are not allowed to see. For example, a CFO may be allowed to see all HR salary data; an HR division administrator may be allowed only to query salary data within his or her division. BI applications must preserve and adhere to application security and operating system security settings as well as any database access control including table, column and row-level security. Ideally, the same ad hoc query issued by two different people should deliver the same result, minus any role-defined security constraints.

 

From the seasoned executive to the customer-facing service representative, a true self-service environment should quickly and easily give all members of the organization immediate access to the information they need to make smarter decisions. It is possible to have a one-size-fits-all pervasive BI offering that satisfies the needs of both power users and nontechnical masses.

 

Organizations must put a strategy in place to put the power of BI into the hands of every member of the organization so that they can rapidly act to maximize business potential.

 

By making BI as easy as searching the web, the ultimate goal of pervasive self-service BI can be achieved: fast, secure, accurate and easy-to-use search for all users, over all BI assets, all without introducing an ongoing training and maintenance burden.

 

Reference:

 

  1. Noetix. “Most Companies Don't Use BI Tools to their Fullest Potential According to Market Survey Sponsored by Noetix.” Business Wire, January 22, 2008. 

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access