kirill_makarov - stock.adobe.com
Automated data storytelling is the future of analytics.
Its rise, meanwhile, could signal the demise of self-service analytics.
That was the premise of a presentation by James Richardson, a research director at Gartner who spoke on Feb. 24 during a virtual conference hosted by data storytelling vendor Narrative Science.
According to Gartner, data storytelling will be the most widespread means of consuming analytics by 2025. In addition, by then a full 75% of data stories will be automatically generated using augmented intelligence and machine learning rather than generated by data analysts.
"It is an inevitability that we move to far higher levels of automation in analytics and that we move away from the current dominant self-service model," Richardson said. "The self-service and visual paradigm that now dominates BI is a limiting factor. It's only as good as the individual's ability to serve themselves."
Depending on the source, it's estimated that only about 30% of employees in most organizations use analytics as part of their jobs. And despite advances in augmented intelligence and vendors' emphasis on ease of use to try to make analytics accessible to more users, that percentage has become stalled.
Data storytelling, however, has the potential to change that. Drastically.
And given the dearth of trained data scientists and data analysts, it's needed.
Data storytelling is simply the translation of data into common language in order to inform the decision-making process. At its core, rather than a spreadsheet full of numbers or chart or graph visualizing those numbers, it's a narrative about the numbers. Narratives, meanwhile, more effectively enable users to absorb and understand information than visualization alone.
Data narratives alone, however, won't alter the analytics landscape. If left to humans, analytics will remain the domain of a small percentage of data scientists and analysts who possess the data literacy skills to interpret data and develop the narratives needed to make data-driven decisions.
It also leaves the decision-making process, according to Richardson, subject to human bias -- and human error.
Data storytelling essentially consists of an analyst orienting themselves with a set of data, applying visualization techniques, reaching a moment of insight by creating a narrative around what they've discovered in the data and sharing that narrative to inform a decision.
That, however, is problematic because it depends entirely on humans discovering and interpreting something and risks the possibility of relying on an unreliable narrator.
"We make mistakes," Richardson said. "Data stories are powerful, but there are some challenges. How do we resolve that? What we can do is apply compute power to this problem."
James RichardsonResearch director, Gartner
Some vendors are now doing just that.
Broad-based BI vendors, meanwhile, are adding automated data storytelling capabilities to their platforms. Yellowfin, for example, now offers Yellowfin Stories, and Tableau offers Explain Data.
Gartner, in fact, rates Yellowfin's data storytelling capabilities highest among BI vendors, giving it a score of 5.0 on a scale of 1 to 5, according to a slide displayed by Richardson. Board International, Domo, Salesforce, SAS, Sisense and Tableau all follow with scores of 4.3.
Those automated data storytelling capabilities, meanwhile, have the potential to enable analytics to crash through that 30% adoption barrier and finally reach a broader audience of business users, according to Richardson.
"The No. 1 metric of success of analytics with ordinary users -- not data scientists or analysts but people who are making decisions inside organizations that don't have those job titles -- is sustained adoption of the analytic technologies that are put in front of them," he said. "If you think about data stories as a trigger to adoption, you can very quickly see ... why this matters."
And why Gartner predicts automated data storytelling will become the most common form of analytics consumption in less than five years.
"We're not always right, but we try," Richardson said.
With automated data storytelling's ascent, he added, will come the fall of self-service analytics.
With automatically generated data stories, there will no longer be a need to lean on the small percentage of people within an organization with the expertise to do similar analysis. Meanwhile, Richardson said machines are actually better at spotting patterns and anomalies in data than humans will ever be. They're also more economically viable.
"Plan for a future where self-service analytics is not the dominant form for most users," Richardson said. "You will keep some analysts and you will keep some self-service, but for most, machines will be doing the heavy lifting."