Photo-K - Fotolia
BI and analytics vendors continue to modernize their offerings with machine learning and AI, increasing their augmented analytics capabilities. Their efforts make it faster and easier for data scientists, analysts and citizen data scientists to get the insights they need. However, trying to choose among vendors can be confusing when their offerings sound -- and often are -- very similar.
Traditionally, BI has been associated with reports and analytics have been associated with dashboards, although the distinction between them is fading as the result of augmented analytics tools.
"Augmented analytics will fundamentally change the user experience for analytics," said Rita Sallam, an analyst and fellow at Gartner. "In fact, the dashboard as the primary mechanism and gateway to insights is likely to decline because when you have insights that are automated in a user context and explained in natural language, you can deliver a very personalized experience to the user."
So, instead of exploring a static dashboard and turning the matter over to an analyst for deeper exploration and an explanation of the findings, augmented analytics "skips to the end of the story" by providing dynamic data stories that explain patterns, drivers, clusters, outliers and correlations. It may also prescribe what the user should do based on the findings.
The major players and how to choose among them
Gartner and Forrester Research declined to recommend one vendor over another in interviews for this story, but their respective reports provide some guidance. Neither of the firms rank vendors on their augmented analytics tools exclusively, although augmented analytics will play a greater role in 2020.
"What we're seeing is that today's analytics and BI are evolving from [the] visual exploration paradigm popularized by Tableau and Tibco Spotfire to one where insights will be more autogenerated, augmented by natural language and delivered to a person more dynamically based on their context, their role and what they've looked at before, what groups they belong to and what exploration they've done in the past," Sallam said.
Some of the vendors Gartner and Forrester have both been tracking, based on Gartner's "2019 Analytics and BI Magic Quadrant Highlights" report and the "Forrester Wave: Enterprise BI Platforms (Vendor-Managed), Q3 2019," report include: Birst (acquired by Infor), Domo, IBM, Information Builders, Looker, Microsoft, Oracle, Salesforce, SAP, Sisense, Tableau Software (acquired by Salesforce) and Tibco Software.
The 2019 Gartner Magic Quadrant also includes Board International, GoodData, Logi Analytics, MicroStrategy, Pyramid Analytics, Qlik, SAS, ThoughtSpot and Yellowfin. The Forrester BI report also includes 1010data and Amazon Web Services, although Forrester VP and principal analyst Boris Evelson said he's also tracking MicroStrategy, OpenStax, Qlik, SAS and Information Builders' WebFocus.
Boris EvelsonVP and principal analyst, Forrester
Evelson said he struggles to differentiate among vendors because if one introduces a capability, all the others follow quickly. He recommends that organizations pay attention to the nontechnology aspects of vendor selection such as their relationship with the vendor, pricing and the availability of resources.
"Don't spend a lot of time looking at each individual feature and capability. Look at all of the adjacent factors," Evelson said. Some vendors provide full-stack capabilities, while others offer a BI layer and rely on partners for the other capabilities. Moving forward, Evelson anticipates more consolidation from vendors.
Other considerations are a company's strategy and what already exists in the organization. If a large enterprise is committed to migrating everything to Microsoft Azure, Evelson said it would be hard to recommend anything other than Microsoft Power BI to them. Similarly, if their enterprise ERP and CRM suites come from SAP or Oracle, then he'd find it hard to recommend anything other than SAP or Oracle tools, respectively.
Be patient with natural language processing
Analytics and BI vendors are providing two types of natural processing capabilities: natural language understanding (NLU) and natural language generation (NLG). NLU is necessary to comprehend a user's typed or spoken queries. NLG explains data visualizations and enables natural language responses to be typed or spoken queries.
Natural language interactions tend to be text-based today. In the future, BI and analytics platforms will have more and better voice capabilities.
A word about automated machine learning
Gartner's Sallam considers automated machine learning "the future of augmented analytics" because machine learning and AI automate feature engineering, as well as aspects of model selection, model management and explainability. Automated machine learning can also identify potential bias in a model and privacy violations.
"You've got disrupters like DataRobot, H20.ai and RapidMiner, but now Microsoft Azure is including [automated machine learning] autoML features or augmented analytics in data science or machine learning platforms," Sallam said. "You see SAS and IBM providing those capabilities."
While automated machine learning helps make data science and machine learning experts more productive, it also helps democratize data science and machine learning so less skilled individuals such as citizen data scientists or analysts can take advantage of some of the same capabilities. In some cases, automated machine learning and augmented analytics capabilities are converging.
"There's a really long tail of vendors [that] focus on making it possible for somebody without any data science or coding skills to upload a data set, select a target they want to predict, and it will spin up data to generate a whole bunch of different feature combinations, try a bunch of different algorithms, validate them to make sure there's no overfitting and rank them by accuracy," said Kjell Carlsson, senior analyst at Forrester. "Some usually give you the option of choosing a performance metric and sometimes they'll let you deploy it as well."
One size does not fit all when it comes to these augmented analytics tools, however. Carlsson mentioned two types of tools, "multimodal predictive analytics and machine learning solutions" that enable data scientists to make predictions more effectively and "automation-focused machine learning solutions" for business users that automate everything.