Business intelligence and analytics platform vendors are now providing augmented analytics capabilities that empower citizen data scientists. Specifically, they use natural language understanding capabilities to enable natural language searches and deliver the results using natural language generation. The result is a "conversation" between the user and the system. Augmented analytics tools also come with pre-built machine learning models to empower any user to do single click forecasts, identify trends and trend reversals, anomalies, outliers -- tasks that in the past required involvement from professional data scientists. In short, the opportunities are many, but many enterprises have a way to go before their businesses are truly "insight-driven."
According to a recent blog by Forrester VP and Principal Analyst Boris Evelson, "Only 7% of firms report advanced, insights-driven practices…less than 20% of all raw business and operational data makes it into analytical databases and applications and only 20% of enterprise knowledge workers who could be leveraging enterprise-grade analytical applications are doing so."
Although the augmented analytics vendors are continuously adding new capabilities to their products, there are some fundamental things that enterprises need to get right.
Why enterprises are struggling
BI and analytics vendors are fiercely competitive, so much so that Evelson sees little differentiation among them. As soon as one vendor introduces a new feature, the others include their own version of it in the next release.
However, the reasons enterprises are not realizing their analytics goals has less to do with technology and more to do with processes and people.
"Technology is only one small part of the overall success equation. It's people, process, data, strategy and how much executives are willing to invest in this," said Evelson. "The reason enterprises still haven't taken full advantage of all the benefits augmented analytics can provide is not the fault of technology by any means."
However, technology also plays a part. For example, although relational and multidimensional databases can answer many questions, graph databases are necessary to understand relationships.
"Once you reach three or four degrees of separation, traditional analytics don't work very well so you have to have something like graph analytics," said Evelson.
He also said a lot of work still needs to be done building a semantic layer on top of analytics.
Further, earlier generation analytics didn't address the I-don't-know-what-I-don't-know dilemma (unknown unknowns).
"Why can't we have a system that runs in the background that can find something interesting, such as an anomaly, outlier, exception or trend reversal without me having to ask for it? That's what we see as a future of augmented analytics" said Evelson. "We see augmented analytics as a push type of analytics whereas [in the past,] I needed to pull information, ask for it." These types of technologies are beginning to pop up. Forrester calls this market subsegment "Anomaly Detection," but it's still early.
What vendors are doing
BI has been associated with reports and analytics with dashboards. However, BI and analytics vendors both provide reports and dashboards. More recently they have added augmented analytics capabilities.
Analytics dashboards provide a way to monitor KPIs. They also provide interactive capabilities such as filter, sort, and pivot so users can view data in different ways.
"The challenge with that is it's extremely manual. [It's also difficult] for somebody who's not a skilled analyst," said Rita Sallam, distinguished VP analyst and Gartner fellow. "We see augmented analytics as an extension of that dashboard paradigm."
Instead of discovering insights through dashboard interaction, augmented analytics automatically generates insights based on users' natural language queries. The charts include narrative text that explains what the chart means. The natural language search and narration capabilities make augmented analytics tools easier to use and more practical for use by "citizen data scientists" who are power users.
Training is important
Although augmented analytics tools abstract the complexity of analytics, the resulting simplicity is not a substitute for analytical thinking. However, most people in lines of business lack formal data analysis training.
"The more you automate, the more you have to train people," said Sallam. "In the past, we used to train people on how to use the tool. Now we have to teach people how to think critically, use the data within their context, and understand the concepts of when something is valid or invalid so they can make assessments about the automated insights that are being [presented] to them."
A broader question is whether the analytics that's being done in a way that aligns with business goals or not because insights have little, if any, value if they don't advance what the company is trying to achieve.
Another important point is that citizen data scientists are not substitutes for data scientists. The two should be working together. To do that, citizen data scientists need to be data literate.
"You can't just have an analyst or even a consumer go off and use insights that haven't been vetted and validated," said Sallam. "Collaboration between the expert data scientists and these emerging citizen data scientists are critical."
Anand Rao, global artificial intelligence lead at multinational professional services network PwC said everyone at his company is being trained on digital technology, which includes data analytics and intelligent automation, because they go hand in hand.
"We don't expect them to become data scientists, Python developers or software developers, but we do want them to know what the system is doing, why it's doing [it], how to ask the right questions and also understand the key risks they should be thinking about when a system makes a recommendation. So not to take everything at face value but be able to prove and see when [a recommendation] is contradictory to your intuition. I think that kind of learning is very much needed across the board."
Augmented analytics tools help organizations become more insight driven, but a tool is just a tool. Enterprises need to do the heavy lift to get data in a state in which it can be used and accessed as appropriate, putting the necessary guardrails around it such as governance and security and training augmented analytics users so they are capable of driving business value from insights.
Follow up with these articles on augmented analytics: