The future of 'pervasive BI'

Many business intelligence (BI) tools present a ‘one-size-fits-all’ solution. But the BI tools of the future might look quite different.

Business intelligence (BI) refers to technology that helps companies make better decisions. Traditionally, BI includes reports, dashboards and scorecards that monitor various parts of the operation. BI enhances business data by turning it into useful information, or builds knowledge by gathering insight about a particular domain - from production to sales, finance to HR.

Enterprises started to adopt BI on a wide scale 25 years ago. The initial motivation was to empower executive managers with current insight about business operations so they could align corporate strategy with employee performance and culture. For a long time, the BI market was dominated by two types of tools: one that generated static reports for the masses and another that used analytics to help technically savvy users slice and dice data in very sophisticated ways.

As the years passed and the value of BI became apparent, industry pundits coined the term ‘pervasive BI’ to describe a more extensive vision. Everyone makes decisions, they reasoned, so why not help them make more informed, fact-based decisions? This philosophy launched a wave of end-user-oriented BI tools. And yet the reality is that BI penetration remains dismally low. Today, less than 30% of potential BI users actually use BI, according to Gartner.

Stuck in the past

One reason for this disparity is that business professionals and BI vendors often cling to yesterday’s perspective that BI is for analysts. Economists call this ‘path dependency’ - following the same old approach instead of switching paths. Every time a BI initiative fails to expand beyond the early-adopter phase (which typically includes the analysts who have selected the solution), they tend to blame it on the usability of the tools. What’s the typical solution? Get new tools, of course, in the hopes of remedying the situation. The success rate does not increase in each cycle - but the frustration certainly does!

Usability is important, but we should also consider what behavioural economists have learned about the human decision-making process. Nobel Prize laureate Daniel Kahneman describes two sides of the brain in his book Thinking, Fast and Slow. One side of the brain makes a decision in a split second; the other side takes time to analyse and ponder the facts. Other behavioural economists have coined the term ‘tacit knowledge’ to describe how professionals use their experience to quickly glean facts and make decisions. Our fast and slow mental systems work together. And so it should be in the enterprise as we set our minds to solving a variety of business problems.

For a similar example in the business world, take a look at labour statistics to see how jobs are allocated. The US workforce includes two million employees with the word ‘analyst’ in their titles. On top of that, there are six million executives, 19 million managers and 70 million operational employees with decision-making authority. This division of labour reflects an important issue in the decision-making potential of most enterprises. The two million analysts spend 90% of their time analysing data. The rest of the employees must make decisions very quickly to expedite the execution of vital tasks.

Clearly analysts need sophisticated BI tools to do their jobs. And as the volume, variety and velocity of data grows, these professionals require progressively more powerful tools for data management, data integration, data quality and data analysis. Some analysts need ad hoc tools to answer routine questions. Others need a general-purpose BI environment to uncover new ideas and create new products. For multivariate datasets, analysts need machine-learning tools that can detect patterns in large volumes of data. And, of course, statisticians need specialised tools to model data and create predictive applications. All of these professionals have different skills and roles within the enterprise. One tool would definitely not fit all.

Consider the software tools that Adobe has brought to the world of publishing. Adobe’s suite of products accommodates a broad division of labour, with photo editing tools to work with individual images, illustration tools for drawing and design tools for laying out entire magazines. Adobe’s success in this market stems, in part, from the interoperability of these tools: a project begun in one of them can be passed on to another.

Similarly, the BI industry should never have attempted to package every conceivable function into one tool set. Far better to create a related family of tools that serves a diverse group of data-management professionals - and, increasingly, rank-and-file business professionals as well.

Keeping it simple

Remember, despite their varying needs, professional analysts comprise a small segment of the decision-making workforce. What about the 100 million working professionals who need information at a glance? We can’t saddle them with unwieldy BI tools that require a learning curve. This busy segment of the workforce is accustomed to simple, intuitive apps that yield answers with a couple of clicks or taps, like the weather app on your phone. Enter a postcode to see the five-day forecast, then tap an icon or two to drill into the details. Whether you want an hour-by-hour breakdown or a moving satellite image, it’s obvious what you are supposed to do.

This is exactly what business professionals need: simple BI apps that can answer basic questions - ideally, with a user interface that has been tailored to the job function at hand. Pair this functionality with an app-store-like portal, where users can assemble their own dashboards from a library of components, and you have a great way to expand an organisation’s decision-making potential.

This vision represents an entirely different mindset from what the BI industry has espoused for most of its history. Rather than waiting for IT to develop dashboards and reports, business users can manage their own content in the portal. They can assemble their own dashboards by dragging and dropping the pertinent components from a repository or choose job and function specifics. This is precisely the type of consumer-oriented approach that can make BI pervasive. Casual business users want to manage their own content via interactive apps that are limited in scope and focused on a specific business problem. Whether they are at the office or on the go, they can call up the right app, interact with it and get the results they need.

Unlike full-fledged BI applications, these apps aren’t designed to manipulate arbitrary data sets. They only include the data that is pertinent to the problem at hand. While dashboards and reports might require you to go back to IT for additions and customisations, these apps let you customise your content with ordinary controls such as drop-down boxes and clickable graphs. This self-service approach permits ordinary people to get answers fast, without calling an analyst or loading up a spreadsheet.

Pervasive BI - revisited

Professional analysts will always want advanced tools. What we are talking about here is empowering the next wave of thinking professionals. Do pilots and surgeons analyse information in Excel when they need to make a split-second decision? Or do they simply glance at equipment consoles to assess the status of current conditions?

Leave the tools to the people who need to pour over the data and provide simple BI apps for everybody else. Toyota calls it ‘visual controls’ (Toyota Way Principle 7) - bits of information, presented in context, that encourage you to act or react.

This brings up another important requirement. These apps shouldn’t just present information. They should offer advice, such as indicating whether or not a condition is acceptable and providing guidance about what to do next. In every business, important events happen during the execution of core operational processes. Yet few companies have the ability to detect and respond to these events the moment they take place. This leaves them unable to act decisively when critical business conditions occur, such as a complaint filed by a large customer, the receipt of a large sales order or the onset of hazardous weather conditions that may delay deliveries.

Such apps bring business intelligence and operational processes together to empower people to react to critical business events immediately. That’s the promise of pervasive BI.

*Dr Rado Kotorov is Vice President of Product Marketing for Information Builders, a BI vendor that helps organisations transform data into business value.