Why Natural Language Processing is NOT the Future of Business Intelligence


For the first time in history, Natural Language Processing (NLP) has become ubiquitous, having found its way into millions of homes, answering billions of questions — just ask Alexa. It’s most certainly a powerful tool, but if you’ve ever asked Alexa a complex question, you’ve likely noticed the limitations. Those same restrictions apply in Business Intelligence (BI).

Gartner predicts that by next yearOpens a new window , 50 percent of analytics queries will either be automatically produced, generated via search or through NLP, but that still leaves 50 percent being generated through other means. While the BI community is bent on cracking the NLP code, in order to save time, money and resources; the technology, as it currently stands, is not mature enough to solve all these problems. Aside from the massive upfront implementation, ongoing maintenance and expenses required to successfully implement NLP, the major shortfall remains in the technology’s ability to handle complex questions.

No matter how mature NLP technology becomes, it will always be dependent on the kinds of questions the data team thought the business team would ask and trained the system to answer. Organizations need a solution that gives people who don’t know SQL and can’t code the ability to securely, safely, and independently explore data without the fear of “breaking” it or accidentally deleting anything. One with a familiar interface, like a spreadsheet, that can support extremely complex questions and that nearly anyone can quickly start using. Business teams have earned the right to access, explore, and analyze the data they contribute to producing and the data that will help them do their jobs better.

Learn More: Here’s Why AIOps is the Magic Bullet for Root Cause AnalysisOpens a new window

The dream of using NLP for business intelligence is the ability to turn any question, borne of natural language, and turn it into a query that can be understood by data warehouses built on SQL. But SQL is already a language that can be learned and utilized. All that NLP will accomplish is to put a shiny new veneer on what is ostensibly a translation tool. One that is designed to mimic natural language requests, which are so often too imperfect to be useful. And as we’ve seen when trying to ask Alexa about anything more complicated than reporting on the day’s weather forecast, NLP comes with a huge potential for error.

Even if we assumed that the perfect NLPOpens a new window translator can be created, it would still be confined to answering the questions it is programmed to answer. Ultimately, you end up back at a programming language, defined entirely by what the developers of that application were able to program. This can be incredibly limiting for business teams looking to find the right piece of data that will help differentiate them from their competitors—and creates feedback loops that inhibit the creative use of the data available to them in the end.

Learn More: How to Turn Failed AI Initiatives into Real Business ValueOpens a new window

True BI innovation comes in the form of an easy-to-use interface that is powerful enough to run complex SQL queries, but flexible enough to meet the needs of anyone looking for answers. That interface must be able to express mountains of detailed and complex data in a form that nearly anyone can understand, manipulate, and analyze. So, what better form can there be than the humble spreadsheet? In its basic form, a spreadsheet allows for the complex collection, manipulation, and analysis of data, in a way that is intuitive and ubiquitous across every industry. Putting an analyst’s capabilities at the fingertips of business and domain experts with little or no understanding of SQL, all without the need to write a line of code.

With all of its subtleties, NLP is advanced, but ultimately limiting tool. An interface that is flexible, intuitive, and powerful like the spreadsheet, when empowered with the querying capabilities of SQL, puts the vast potential of data in the hands of those who can make the most use of it. This is the true democratization of data.

Let us know if you liked this article on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We would love to hear from you!