While analytics has undeniable value for enterprises that use it wisely, bad implementations are leaving some early adopters burned.
Nearly half of companies with big-data projects aren't getting the value they anticipated, according to a study from a Wikibon analyst. More specifically: Some 46 percent of big-data practitioners report they've only realized partial value from big-data deployments, said Wikibon analyst Jeff Kelly in a report: "Enterprises Struggling to Derive Maximum Value from Big Data."
The report adds: "An unfortunate 2% declared their Big Data deployments total failures, with no value achieved."
"The first common scenario is that enterprises invest in Big Data technology such as Hadoop without specific and measurable business applications tied to the projects," Kelly wrote. IT mostly drives these cases. They collect big volumes of data in Hadoop. Sometimes that data is made available to data scientists and business analysts for exploratory analysis. But otherwise the data sits underutilized.
Sometimes, enterprises successfully test big-data pilot projects only to find they don't have enough skilled practitioners to support production deployments, Kelly said. These practitioners include administrators, developers, and data scientists.
Expectations are a stark contrast to reality. Enterprises expect to get $3 to $4 for every $1 invested in big-data over the next three to five years. In reality, most enterprises just get $.55 on the dollar, Kelly said.
But some enterprises succeed with big-data. These projects are generally not initiated by IT but by line-of-business departments, often marketing, Kelly said. They focus on small but strategic use cases. Practitioners assess the skill level inside the enterprise and often engage outside professional services organizations to fill talent gaps and keep systems running optimally.
"The cost of engaging Big Data services will most often be more than offset by the increase in value achieved by successful Big Data projects," Kelly said. "Professional services organizations are also useful at identifying initial Big Data use cases to that result in value in-and-of-themselves but also serve as proof-points for driving future projects."
These problems and more have led to "bursting of the big data bubble," writes Cathy O'Neil on the blog mathbabe.
Enterprises look to companies like Google, which have been wildly successful with big-data, and think they'll see the same success. But that's not going to happen, O'Neil said:
Most companies donít have the data that Google has, and can never hope to cash in on stuff at the scale of the ad traffic that Google sees. Even so, there are lots of smaller but real gains that lots of companies -- but not all -- could potentially realize if they collected the right kind of data and had good data people helping them.
Other problems outlined by O'Neil: Business managers are disconnected from data scientists, there are no standards for who can call himself a data scientist, and it's hard to assess qualifications.
But O'Neil is optimistic:
My forecast is that, once the hype wave of big data is dead and gone, there will emerge reasonable standards of what a data scientist should actually be able to do, and moreover a standard of when and how to hire a good one. Itíll be a rubric, and possibly some tests, of both problem solving and communication.
I agree. Companies are seeing real benefits from analytics. Unfortunately, new, useful technologies are often overhyped, leading early adopters to get burned. But in the long run, analytics is here to stay.
But maybe not the phrase "big-data." Like "information superhighway" and "dotcom" before it, this piece of jargon seems destined to become a fatal victim of the hype.
— Mitch Wagner , Editor in Chief, Internet Evolution