Page 3 of 3
Yellowfin Displays Adaptability
Glen Rabie, CEO of Australian business intelligence (BI) firm Yellowfin Inc., says he used to enjoy writing reports. He still does.
A decade ago, when Rabie and his team were designing Yellowfin, reporting was the overwhelming focus of BI. To most enterprises, "BI" meant reports.
That's one reason the former Business Objects SA spent almost $1 billion for the former Crystal Decisions Inc. back in 2003, and why -- just weeks later -- the former Hyperion Solutions Inc. paid several hundred million dollars for reporting specialist Brio Software Inc. It also explains why Microsoft Corp. introduced its own reporting offering -- SQL Server Reporting Services (SSRS) -- a few months later.
It's no wonder, then, that Rabie and his team designed Yellowfin expecting that it would function as a reporting platform. Luckily for them, they made several core design decisions that helped to make it a surprisingly adaptable platform.
In the last few years, BI has quickly moved beyond reporting -- and so did Yellowfin.
"I used to be a person who loved writing reports. I still am. In our original vision we assumed everybody would write reports," Rabie explains.
As Yellowfin and (to a large degree) the rest of the BI industry discovered, users as a whole are more interested in consuming than in creating information.
"What they really want is to consume a really easy-to-use dashboard that just tells them [what's going on]," says Rabie. "That [knowledge] ... really drove our product [development] into a much more collaborative direction."
Fortuitous Mistakes
Collaboration came later, of course. First, the Yellowfin team had to make a few "fortuitous" mistakes. The biggest of these, Rabie says, was a decision to focus on delivering a completely browser-based BI experience -- in 2003, back before Software-as-a-Service (SaaS) had really caught fire.
The impetus was Rabie's experience with a large financial institution, which spent millions of dollars as part of an effort to deploy a then-typical on-premises BI offering to a few hundred users. Rabie says he was determined to make Yellowfin easy to deploy to client desktops, and to eliminate -- as much as was possible -- the implementation or connectivity issues that tended to complicate or even imperil BI implementations. At the time, Rabie says, he imagined that most of Yellowfin's business would be in the form of conventional hosted SaaS, a la Salesforce.com.
This turned out not to be the case, however.
"We went with purely a thin-client model that said everything was going to be delivered by the browser. We baked into [Yellowfin] right from the start ... all of the multi-tenancy requirements that we needed to have. [Today,] we supply a lot of SaaS vendors with Yellowfin. All of that multi-tenancy logic was built in from day one," he explains.
"I would call that our biggest fortuitous mistake. It was too early for SaaS business intelligence, at that stage. Even now, a pure generic SaaS business intelligence model is difficult. That mistake, thinking we could deploy it in the cloud [i.e., as hosted] ... but realizing that everybody wants it internally, that was great for us."
Another "fortuitous" development was the "whacky" approach that Yellowfin took to reporting, Rabie notes. "Traditional report definitions were done on a document basis, so they were stored in an XML format or [in] some [proprietary] artifact like that. We took this whacky approach of actually building a data model for [Yellowfin's report definitions]," he explains. "That meant ... that we could continue to extend the data model around the core product."
This approach has several advantages, he maintains, not least of which is that it enables Yellowfin to non-disruptively introduce changes, enhancements, or new features. "If you started [from] day one and had one XML file [for report definitions], which most people do, it is painfully difficult to move from that and wrap all of that other [i.e., new or additional] stuff around it. That was one of those cute things that allowed us to scale. We didn't set out [from] day one to build this enterprise platform, we just got there as our clients asked for more and more. We just wrapped in new functionality. We still do that."
Yellowfin uses this same data model to generate reports, dashboards, and other views, he says. "Our Dashboard Builder is hierarchical, so you build a report, and you can use that report in any [device or context]. In most other business intelligence products, the dashboard is something that you build completely separately from the reports that you actually write. It's a subtle but hugely different thing. You build that dashboard by rebuilding all of the logic that you had in the report, specifically with our dashboard builder."
Yellowfin's first mobile interfaces were "pretty ordinary," Rabie concedes, but the company has since tweaked its Dashboard Builder component to generate optimized mobile dashboards. "If you build that dashboard in your browser, [we say] let's just extend that to a mobile device -- so we will change the way you interact with it [so that it's suitable for a mobile context], but all of the content will be the same. These are the subtle but really important things that make enterprise deployments possible," he indicates.
"We have a really high degree of personalization at the dashboard level, and although you can share common content -- the whole sales team could share a set of sales tabs -- you could really personalize [it] ... to your experience and your particular needs in your organization. We change [the presentation] depending on the device you're using, ... not the underlying content."
Evolving to Collaboration
Yellowfin's ongoing focus is collaboration, says Rabie. To that end, it's evolving its BI platform with a collaboration-first emphasis. This emphasis determines how it incorporates or addresses new features, functionality, or requirements, he says. A good example of this is Yellowfin's approach to social BI.
"Our job as the vendor is to make products that are engaging and that people get excited by. Our goal is to make [BI] something [with which] you want to interact," he explains. "For example, the way we do discussions ... you can follow content to see whether people are relating to [that] content. You can get new content [on this basis]. It's not social as in 'social,' [although] we are adding new features like following people."
"To me, [social] comes under the big umbrella [of] collaboration. How do you collaborate on reports [or] share content with other people? Why do you find things insightful? What we've started to do is to bubble more of that stuff to the surface to make it far more interactive around sharing and content."
About the Author
Stephen Swoyer is a technology writer with 20 years of experience. His writing has focused on business intelligence, data warehousing, and analytics for almost 15 years. Swoyer has an abiding interest in tech, but he’s particularly intrigued by the thorny people and process problems technology vendors never, ever want to talk about. You can contact him at
[email protected].