Cheezburger Avoids BI ‘Fails’ By Sipping Data Slowly

Getting value from big data is everyone’s problem, even LOLCats.

If you think staring at LOLCats or the FAIL Blog is wasting time at work, try telling that to Notorious BIT—that’s the Business Intelligence Team, a three-person delta force in charge of analytics at Cheezburger, the online humor network. With more than 30,000 websites in its network, Cheezburger can collect a tremendous amount of data, according to an article recently posted by Software Advice.

As with many big data efforts, collecting data wasn’t the problem—getting value from it was. Then, once a big-data solution was in hand, that common first-order problem was quickly compounded by a second-order problem—addiction to reporting. Getting its hands on the business discovery tool QlikView, the team’s standard process was to start guzzling data and generating reports. They’d spend hours analyzing one to three months’ worth of data, only to find they still didn’t have what they were looking for.

Loren Bast, director of Business Intelligence at Cheezburger was quoted as saying, “The problem that we thought we were solving was the ‘Big’ problem. But the problem that we really should have been solving first was the ‘Data’ problem.” The team had to figure out how to stop the madness and identify analyses that would produce value.

The Key Lay in Doing a Few Things

  • Spending more time at the point of collection ensuring data validity. As we pointed out in “What BI Stays and What Goes?” we are not about to be relieved of the task of data cleansing just because the nature of the data we ingest is changing. The BIT took the time to validate new data sets against legacy systems for accuracy before taking the step of automating reporting.
  • Always asking if something valuable to the business would come out of the report before generating it. If the answer is “no,” then there’s no report.
  • Choosing metrics that are actionable, without agonizing over them. It turns out the easiest key performance indicators (KPIs) are actually the most insightful at the beginning.
  • Building buy-in with a broader audience. Writing “Insanely Insightful Commentary” on top of its daily BI reports, focusing less on reporting and more on inviting discussion of potential issues raised by the reports, which, in turn, would more clearly identify issues that mattered to the business.

Even if you’ve been reprimanded at work for “wasting time” staring at LOLCats, you could choose to find a perverse comfort from knowing the people on the other side of the screen were spinning their wheels staring at you staring at LOLCats. But the real lesson here is broadly applicable to all kinds of businesses in the throes of big data paralysis. Periodically evaluating both tools and practices can cut down significantly on BI “Fails.”