• Data and analytics have historically belonged to the technological elite, those able to understand the subtleties of inner vs. outer database joins or those willing to learn the difference between a scatter plot and histogram.
• That’s about to change, thanks to a groundswell of innovations emerging from analytics vendors which promise to make data analysis as easy as asking Google Assistant for the day’s weather forecast.
Here’s a very simple proposition to consider: What if we don’t really need citizen data scientists or business-savvy data specialists to realize the current dream shared by most in the enterprise data and analytics marketplace — that dream being the true democratization of data and data-driven insights.
What if instead of asking data specialist to put together a static report — an often iterative and time consuming process — anyone in an organization could type out the simple question, “What were this quarter’s sales numbers?” and get back a timely and accurate data visualization of that quarter’s sales numbers.
This sounds simple, but in the realm of data visualization and discovery, that simple question would typically require the selection of the correct data set (often from among hundreds of available databases and spreadsheets), the selection the most appropriate chart type and then the proper filtering for the desired date range.
And what happens when the user wants to then refine that query, or dig a little deeper into the numbers themselves? Historically, that has necessitated a further conversation with a data and analytics professional and further time wasted in arriving at the desired insight. In response, most analytics vendors have tried to make it easier for non-data professionals to drag-and-drop data sets and chart types, often using guided step-by-step advice. But that still requires specialized training centralized support.
The application of artificial intelligence (AI) capabilities such as natural language processing (NLP) promises to do away with both sides of this problem (specialized knowledge and central administration). So far, this has fallen to AI and analytics market leaders Microsoft, SAP, and IBM, which individually have made several pointed forays into the realm of AI-driven data democratization. Now pure-play analytics vendors are getting into the act. This week at its 11th annual user conference, Tableau Software launched a new capability branded Ask Data, which directly equips any user with the ability to ask questions like “what were this quarter’s sales numbers?”
Built using technology acquired from ClearGraph back in 2017 (technically not pure AI but NLP plus some statistical reasoning), Ask Data converts natural language into the machine query language used by Tableau to retrieve the right data and display a clear answer in an interactive chart. What makes this conversational, however, isn’t the initial question itself but rather the iterative refinement of that question in natural language. Using the above example, a user could refine his/her query by, for instance, adding “but only for the northwest region compared with this same quarter last year.” Tableau will then gather up the necessary data tables to factor in those new additions (regions and past performance) all without forcing the user to start over or seek outside help.
The work going on behind the scenes with Ask Data is decidedly non-trivial and still demands hands-on expertise. Assumptions regarding the meaning of statements like “sales numbers” demand that the system take into account who’s asking the question and how the business defines such terms, for example. At some point well before the user asks a question, data practitioners and business owners will need to define a common language (as it were). In other words, solutions like Ask Data are not a plug-and-play affair; they demand that a company have or get its act together on the data model and semantics front in order to leverage such solutions most effectively.
What does the future hold? For Tableau one route forward will be to use NLP to do things like edit charts, dashboards, queries, etc. Another one area of investment firmly on its development roadmap is the additional of AI-driven digital assistants like Amazon Alexa, Apple Siri, or Google Assistant. With one of these functioning as a user interface, it’s not a huge leap to imagine a scenario where a busy executive could carry on a conversation via his/her telephone, not with an assistant but instead with the business itself. “Hey Google, should I carry an umbrella today and how did we do on last quarter’s umbrella sales in New Orleans?”