Big BI Lessons from Cheezburger’s Notorious B.I.T.


While everyone argues about how to concretely define “Big Data” and what businesses should do about their “Big Data problems,” there’s one thing we can all agree on: we sure do love talking about it. According to Google Trends, search volume for the phrase “Big Data” has exploded since 2011.

But while it’s inevitable that the Trough of Disillusionment will follow the Big Data hype, so is the inevitability of failure for many new users of business intelligence (BI) applications. In fact, Gartner found that between 70 and 80 percent of enterprise BI projects fail and predicted that fewer than 30 percent of BI projects will meet business’ predetermined objectives by 2014.

Until recently, online humor network Cheezburger may have been trending toward those outcomes. In 2011, the company–which operates a number of popular websites, such as I Can Has Cheezburger, LOLcats, and FAIL Blog–found itself unable to keep up with the data it was producing. In need of an application to process its visitor data in near real-time, they decided to ditch their internally-developed analytics program and implemented QlikView.

But the team that selected the application soon realized it was looking at its Big Data problems in the wrong way.

“The problem that we thought we were solving was the ‘Big’ problem,” says Loren Bast, Director of Business Intelligence at Cheezburger and the individual who oversaw the implementation of the application. “But the problem that we really should have been solving first was the ‘Data’ problem.”

Cheezburger’s three-person business intelligence team–which Bast coined the “Notorious B.I.T.,” after the famed rapper Notorious B.I.G.–needed to re-envision some of its standing processes to provide the best possible analytical offering to the organization. They did so by ensuring it could stand behind its data analysis, identifying analyses with the most value and creating a platform for discussion within Cheezburger to reflect on larger company trends.

Sweat the Small Stuff with Big Data

During the first few months of using QlikView, Cheezburger fed all their data into the system and immediately began reporting. The ability to analyze any demographic against a number of measures in real-time was exciting–until they realized incomplete datasets, invalid data and miniscule information discrepancies led to inaccurate analytics.

“We would spend a ton of time analyzing one month or three months’ worth of data, only to realize the data we were looking at wasn’t exactly what we were looking for,” says Bast.

Today, new data are introduced gradually and compared against a legacy system for accuracy before reporting is automated. “We capture only a few feeds, make sure they’re right, and then we build upon that,” says Bast.

Ensuring data is valid, accurate and complete should be one of the first steps in any new BI project, but it’s often not given enough attention, says Gurpreet Singh, IT Director at AccoStar Technologies.

“It is not just the diversity of sources that introduces complexity, but also the typical lack of data integrity that makes it essential to give more emphasis to this phase,” says Singh. For larger companies with hundreds or thousands of data sources, he suggests identifying milestones and individual budgets for each data extraction to ensure this process receives the attention it deserves.

Identify Analyses with Value

When first utilizing the new BI application, it was easy for the Notorious B.I.T. to get caught up looking at too many groups of metrics. Aligning the right key performance indicators (KPIs) with objectives wasn’t always at the forefront of the discussion, but Bast realized this discussion needed to happen sooner and was worth the time spent.

Bast says their initial approach when using QlikView was akin to: “Track everything and figure out what to do with it later.” But this led to reporting that just didn’t need to be done, and took away resources to focus on “analyses with value.” Today, the Notorious B.I.T. takes a much more consultative role with the teams for which it creates reports.

“Now, we always ask if something valuable to the business is going to come out of the report,” says Bast. “If we can’t agree on a clear answer to that question, we just won’t build the report.”

In addition to selecting metrics that are aligned with a team’s objectives, Bast emphasizes the importance of choosing metrics that are actionable. But while this is important, companies shouldn’t spend too much time selecting KPI A over KPI B.

“Businesses don’t need to stress over identifying what the best key performance indicators may be,” explains David Abramson, Product Manager at LogiXML. Instead, he advises organizations to select a reasonable group composed of the most obvious metrics, and then build from there. “If it’s hard to measure, it’ll be hard to find value. The easiest KPIs to measure often provide the most insight in the beginning,” says Abramson.

Increase Buy-In with “Insanely Insightful Commentary”

The last change came about quickly and unexpectedly–and changed how the BI team fit into the Cheezburger structure.

A year ago, on a whim, Bast added an additional commentary to its daily BI report that is sent out to the company. Titled “Insanely Insightful Commentary,” the discussion focused less on reporting and more about discussing potential issues and inviting other members of the team to share its thoughts. This was an important move for an organization that’s acquired almost 30,000 websites in its network over the last few years, because the little things can often go unnoticed.

“It sets the rhythm for us,” says Bast. “Opening up the conversation gets the most productive chatter going.” He explains that daily email strings will quickly turn into long threads that will point both product teams and the BI team in new directions.

The coupling of data analysis and commentary provides a strong footing for the BI team within the organization.

“We now consider ourselves a service group,” explains Bast. “We’re sort-of the Seers of the company.”

Do you know of an organization that has re-evaluated the use of its BI tools and seen success? Please leave a note in the comments below.

Share this post:  
Michael Koploy

About the Author

Since joining Software Advice in 2011, Michael Koploy's work has been cited in a variety of online publications, including ReadWrite, O'Reilly, The New York Times and SYS-CON. Michael manages content related to the Business Intelligence (BI) market for Software Advice, writing on BI tools and applications, (big) data-related news and industry trends.

Connect with Michael Koploy via: 
Email  | Google+