The business intelligence (BI) assembly line is broken, with adoption or utilization rates of only 30 percent in a typical organization, according to Gartner[1]. These adoption rates include all the users of the BI system – administrators who manage the system, analysts who build reports, and business users who consume reports for better decision making. This underutilization ignores employees outside the BI system who could be using valuable data to make better decisions. No one can argue that business performance and innovation are negatively impacted when data assets are not fully leveraged for decision making.

How does this BI assembly line contribute to poor adoption? Today, the BI assembly line is highly dependent on several job roles with different skill sets working together to deliver a report or analytic dashboard to a business user to facilitate better decision making. This is not only a lengthy and expensive process, but the BI assembly line also may end up delivering a report to the business user that is incomplete or difficult to interpret because the assembly workers did not fully understand the business. For example, it is not uncommon for a business user to ask more questions outside the scope of a delivered report. When this happens, the business user is left frustrated and requests more data and analysis, kicking off the BI assembly line again.

Figure 1: Today’s BI System: Asking a Business Question Involves Many Roles and Tasks

This assembly line approach not only hinders BI adoption, due to unhappy business users, but it’s not scalable in today’s environment, given the exponential growth in data volume and complexity, and the fact that all organizations need actionable insights faster than ever to stay competitive. The mainstream arrival of AI using machine learning automation promises to boost all kinds of software capabilities, particularly BI. This will enable systems, which know what data to acquire, process and analyze, to deliver the best possible actionable insight to the business user directly. No longer will the business user need to slice and dice the data or ask for more data to answer a business question. As a result, AI has the potential to modernize the BI assembly line by giving business users the answers they need quickly, broadening the reach and value of BI and analytics.

Figure 2: AI Enabled BI: System generates the answer

According to McKinsey[2], technology companies can expect as much as a 10 percent increase in revenues using AI to strengthen product portfolios. How does this apply to a BI technology solution? Because AI predicts the next best step or action to take, not only is there automation during the BI assembly line, but human bias can be removed to deliver the best possible actionable insight. For BI vendors, this means AI enabled BI will propel customers stuck in the “Access and Reporting” phase to the “Analytics” phase.

Figure 3: “Business Intelligence and Analytics” of Davenport and Harris’ Competing on Analytics.

Both AI and BI systems use machine learning to make predictions. Therefore, this synergy makes the BI industry a prime candidate to use and apply AI successfully. Today, BI vendors are already introducing AI capabilities in their products for data preparation, data discovery, and data science. As a result, customers are seeing improvements in quality and time to insights, because the dependencies across the BI assembly line of IT, data engineers, BI administrators, analysts, and data scientists are reduced.

Read Gartner’s latest research, “Doing Machine Learning Without Hiring (More) Data Scientists,” which carves out four recommendations for organizations wishing to launch data science initiatives.

With such innovation today and more to come, this begs the question: Will the BI industry be radically transformed to modernize the BI assembly line, with AI providing the best possible answer to a business question?

Figure 4: AI Enabled BI: Business user asks a question to the system

To better understand the impact of AI, let’s look at how BI tools have evolved over time to address data and analytic challenges, where they fall short, and what the potential is for AI to finally solve the age-old BI problem: time to actionable insight.

The root of the BI problem hasn’t changed.

In 1865, Richard Millar Devens used the phrase “Business Intelligence” (BI) to describe how a banker profited from information by gathering and acting on it before his competition. Business intelligence, as it is understood today, uses technology to gather and analyze data, translate it into useful information, and act on it “before the competition.”

Almost 100 years later, in 1970, Edgar Codd recognized there was a problem with BI in that only individuals with extremely specialized skills could use business intelligence to translate data into usable information. Data from multiple sources was normally stored in silos, and research was typically presented in a fragmented, disjointed report that was open to interpretation. He published a paper proposing the development of a “relational database model” to address this issue.

Unfortunately, it’s now 2018 and the same problems still exists. BI tools still follow a reporting paradigm where a skilled analyst must fully understand the business to assemble the right reports and dashboards that will be valuable to the business decision maker. And the business decision maker often needs more information not found in the prepared reports or ends up spending too much time analyzing and interpreting the information for decision making. Here is a look at BI disruptors, over time, which attempt to make data and analytics more usable, timely and valuable.

Figure 5: BI and Analytics, over time, attempting to increase the reach of actionable insight

Will AI address these BI limitations, transforming the BI industry unlike any other prior disruption?

My take: Although AI has the potential for eliminating, or at least greatly reducing, manual tasks by intelligently and automatically giving the business user accurate and actionable insights, we are still at the early stages of complete automation. Also, humans are curious and still want to know the “why” of a prediction, which AI may not necessarily provide.

I also believe machines cannot replace all use cases and must keep humans in the loop due to potential machine errors. For example, a false positive or false negative for cancer diagnosis can result in tremendous emotional distress, additional unnecessary tests, and even death. For clinical use cases, AI can be used as an effective tool to augment human decision making or guide a physician in diagnosis, instead of a machine providing an answer.

Figure 6: Amazon Machine Learning to build and deploy Predictive Models

In summary, AI is not about man battling it out with machines, but rather man working with machines to enable a new level of BI and analytics. By combining BI with the best AI, vendors have an opportunity to guide users more intelligently and provide them with faster and better actionable insights.

Mona Patel works in Birst’s Product Strategy team. With more than 20 years of experience building analytic solutions at The Department of Water and Power, Air Touch Communications, Oracle, MicroStrategy, EMC and IBM, Mona is now growing her career at Birst. Mona received her Bachelor of Science degree in Electrical Engineering from UCLA.

[1] https://www.gartner.com/doc/3753469?ref=ggrec&refval=3803464

[2] https://www.mckinsey.com/featured-insights/artificial-intelligence/notes-from-the-ai-frontier-applications-and-value-of-deep-learning