Subscribe / Unsubscribe Enewsletters | Login | Register

Pencil Banner

Big Data jumps to the cloud

Maria Korolov | July 9, 2013
Big Data-as-a-Service offers quick, inexpensive, targeted analytics

Though primarily focused on sales and marketing, is rapidly growing into a more general-purpose business platform in the cloud, and the availability of APIs allows for interoperability with both internal systems and other vendors, reducing the silo effect.

Spooky action in an instant
Most Big Data analytics projects aggregate data and then answer questions based on that data. Some automatically send answers to employees who can do something with those insights. But the latest evolution of this technology is to go one step further, and act on those recommendations.

"That's one trend I'm seeing," says Fern Halper, director of TDWI Research, a Eugene, Oregon-based research and education institute focusing on data analytics. "A lot of vendors are talking about 'insight to action' and providing the actionable part."

For example, an analysis platform can track real-time Tweets about a company, decide which ones need immediate attention, and forward them along with a recommendation for specific action to the right person to deal with them.

And some vendors are cutting out the human component altogether. For example, Sailthru's platform can be used to change website pages on the fly, to show visitors content that they haven't seen yet, or special offers tailored specifically to their tastes.

"If you look at T-shirts and buy the red version, we know that the first image of a piece of clothing should be the red option if we have one," says Sailthru CEO Neil Capel.

There is still a role for humans in the Big Data picture, however. "You still need to apply critical thinking to what is coming out of this thing," says TDWI's Halper. "Does it make sense, what it's telling you, does it make sense for the business?"

Say, for example, a company is looking for information about social sentiment. The vendor could be processing trillions of data points, but only a handful might be relevant to that company and even fewer of them have information about gender or geographic location associated with them, not enough to reach a meaningful conclusion about how men and women in particular locations view the company.

"With a lot of these tools, you only have a 50-50 chance of getting the sentiment right," she says. "If also depends on who you're going with if you pay peanuts, you get monkeys."


Previous Page  1  2  3  4  5 

Sign up for Computerworld eNewsletters.