kentoh - Fotolia
2019 BI trends included reality check for AI, consolidation
AI features in BI products improved in 2019, but not at the pace predicted before the year began. Instead, advancement was incremental and consolidation was a dominant theme.
The BI trends that emerged in 2019 weren't necessarily what was most expected.
A year ago, one of the dominant BI trends for 2019 was predicted to be an augmented intelligence takeover.
One vendor, ThoughtSpot, wrote that AI in BI will go mainstream and that self-driving analytics will emerge. Another, Tableau, predicted the rise of explainable AI and that advances in natural language processing would enable users to converse with their data. Meanwhile, the CEO of one more, Yellowfin, wrote that natural language queries would supplant text search interfaces.
The introduction and advancements of AI features was indeed one of the significant BI trends of 2019. Progress was most definitely made. ThoughtSpot, for one, introduced new machine learning capabilities, as did Tibco.
But the progress wasn't quite as spectacular as predicted.
"I think the expectations were a bit too lofty," said Mike Leone, an analyst at Enterprise Strategy Group. "We saw numerous announcements from BI vendors that looked to incorporate AI into their platforms, but adoption is lagging behind quite a bit. Organizations want to use AI to better do their jobs, but there's still a bit of a learning curve and a need to address the simplicity of using the technology for less technical or non-advanced users in these systems."
Similarly, Dan Sommer, global market intelligence lead at Qlik, said that while the advancement of AI capabilities was one of the BI trends in 2019, it was modest.
Dan SommerGlobal market intelligence lead, Qlik
"AI is finding its way into use cases, but in very specific scenarios," he said. "We're still waiting to see the impact of more general AI."
Beyond the incremental advancement of AI capabilities, vendor consolidation was one of the significant BI trends of 2019. Others were the introduction of low-code/no-code tools for application developers, improved mobile apps from BI vendors, and perhaps the demise of Hadoop.
Three weeks into 2019, Qlik acquired CrunchBot and Crunch Data to usher in one of the main BI trends of 2019. One month later, it purchased Attunity.
In the months that followed, Alteryx bought ClearStory Data and Sisense acquired Periscope. Then on June 6, Google bought Looker for $2.6 billion. That same day, Logi Analytics acquired Zoomdata. And just four days after that, Salesforce purchased Tableau for $15.7 billion.
The first six months of 2019 ushered in a wave of consolidation not seen in the BI space since 2007, when IBM acquired Cognos, Oracle bought Hyperion and SAP purchased BusinessObjects.
The impetus for consolidation was the recognition of the growing importance of business intelligence, according to Donald Farmer, principal at TreeHive Strategy.
"What drove consolidation was the commoditization of analytics," he said. "The pressure on pure-play analytics companies has been great, and there are some companies that are naturally struggling and others that are pre-struggling and could see the writing on the wall. But the positive is the drive toward analytics everywhere, the need to get analytics into every part of the business."
While consolidation was one of the significant BI trends throughout the first half of 2019 -- at least among major vendors – it slowed during the second half of the year. That does not, however, mean the wave has subsided.
"The low-hanging fruit has been taken, but I'm sure there will be another round," Farmer said.
A new audience
Software vendors who market far more than merely BI platforms have provided low-code and no-code tools for developers for some time now. Salesforce, for example, debuted Salesforce Lightning in 2014. Microsoft answered with PowerApps two years later.
But one of the BI trends to emerge late in 2019 was the introduction of low-code and no-code tools for developers from vendors specializing in business intelligence platforms.
Looker unveiled Looker 7, including a development framework and an in-product marketplace for add-ons, in early November. The same week, Yellowfin rolled out Yellowfin 9. The release included features called Dashboard Canvas and Dashboard Code Mode, which enable developers to customize their organizations' applications.
Sisense then released an update with embedded capabilities to help customers create enterprise-grade applications, and Alteryx's latest version included an application programming interface that connects Alteryx Designer with Tableau Hyper and allows users to easily read, write and transform Hyper files.
"I think, especially over the last few months, we're seeing a much heavier focus on developer enablement," Leone said. "The idea of low-code/no-code to efficiently build workflows, apply automation and incorporate insights into modern applications is really being emphasized."
The motivation behind the tools isn't to replace developers. Instead, it's to free them from some of the cumbersome coding work it takes to build customized applications. It's also to enable business users to create rudimentary applications, leading to citizen development much the way advances in the ease-of-use of analytics tools has led to the rise of citizen data scientists.
"It's not that developers don't want to write code or can't write code," Leone said. "It's all about efficiency and not having to re-create the wheel."
BI vendors have historically struggled to develop effective mobile apps. Too often, they tried to re-create their desktop dashboards on mobile screens, but the screen size of phones in particular and tablets to a degree made it difficult to consume data visualizations.
The small screen size proved a big hurdle for mobile BI apps.
In 2019, however, one of the BI trends was that vendors began to figure out how to turn mobile devices to their advantage.
The ones now having success -- and the rare ones who had success prior to 2019 -- view mobile differently than they do desktop applications. They recognize the limitations of phones and tablets, as well as the strengths of the devices.
"We saw an emergence of a new style of mobile apps being based on being able to take action -- Yellowfin, Microsoft, Domo," Farmer said. "They're action-focused mobile apps."
MicroStrategy is one vendor that has long invested in its mobile app, first introducing one in 2009. And one thing it does is take advantage of the mobile aspect of mobile devices. Given the app's AI and machine learning capabilities, it's able to provide users with information cards as they move around.
Yellowfin, as mentioned by Farmer, is another vendor figuring out how to present BI in a mobile format.
After struggling for a decade following the release of its first mobile app, the vendor overhauled its mobile strategy and unveiled a new app in September. Rather than mimic dashboards, it presents information via a timeline feed similar to Facebook and Twitter.
BI trends come, and BI trends go. And just as AI and developer tools are gaining popularity and proficiency, Hadoop's time might be up.
"Hadoop ended last year," Sommer said.
Hadoop was created in 2005. It allowed organizations to store and access large amounts of data, and for data scientists to access that data and structure it as needed. In recent years, however, cloud data warehouses such as Amazon Redshift and Snowflake -- among others -- have lessened the need for Hadoop.
In fact, the idea of big data -- the problem Hadoop sought to solve -- may not exist anymore, according to Sommer.
"There's a notion that big data now is just data, and next is wide data," he said. "Big data is just what you can't achieve with your current infrastructure, but with cloud storage that restraint is now gone and you can always add more."
Expanding on the notion of wide data now being an issue rather than big data, Sommer said that data is now fragmented, coming in from different sources at different speeds, and the problem that needs solving is how to bring it all together.
"New data catalogs help pull multiple data sets together," he said.
As evidence of Hadoop's perilous position, the three main Hadoop vendors -- Cloudera, MapR Technologies and Hortonworks -- are in a state of upheaval. Cloudera and Hortonworks wound up merging early in 2019, and MapR, after seeking a buyer, was acquired by Hewlett Packard Enterprises.
"A negative BI trend is that Hadoop is over -- it's done, put a fork in it," Farmer said. "It's had an interesting effect and people are moving toward a different type of data lake architecture -- a hybrid data lake/warehouse … that enables seamless operation between the two."
The same can be said for 2019: It's done, put a fork in it.