1.9.13

Big Data is Already Over - Get Ready for Little Data | VC1212

Big Data is Already Over - Get Ready for Little Data | VC1212

Related Reading:
http://slashdot.org/topic/bi/big-data-hype-is-imploding-gartner-analyst/
http://readwrite.com/2013/01/24/big-data-overhyped-and-overpaid
http://qz.com/81661/most-data-isnt-big-and-businesses-are-wasting-money-pretending-it-is/

BigDataBigBuildingsIt’s all the rage to talk about attacking huge datasets with algorithms in the hopes of mining valuable patterns. Sorry to disappoint you, but it’s safe to say we are deep into the back 9 on the Big Data hype cycle.

No, we’re not going back to hunches and gut feel, but the leading edge of analytic thinking has shifted to the quieter area of what is being called “Little Data.”

What if rather than crunching 3 billion data points in hopes of finding a pattern, the right approach is to find the right few pieces of data to make your decision? Here are a few examples:
  • Fast Food: Burger King noticed that last month, two of their customers purchased 5 medium sodas and 3 orders of fries. So with this insight, they launched a national roll-out of the “Five + Three-er” bundle for $4.99 – which is, you guessed it, 5 sodas and 3 fries. The roll-out will cost $10 million, not the kind of coin you want to spend without the data to back it up.
  • Retail: Wal-Mart saw three instances of no customers showing up between 2:00 and 3:00pm in two of its stores, so it is piloting a nationwide shutdown of its 3000 stores at that time. Again, a quick “intuitive” look at the data would indicate that they are putting over $500 million in revenue at risk, but “Little Data” analysis uncovered this opportunity to save operating costs.
  • E-commerce: Amazon noticed that a customer looking at a Star Wars Lego set then surfed over to a book on wedding floral arrangements. They are dedicating a home page banner promotion to similar bundles. Again, it’s counterintuitive, and the response rate for the first two weeks was 0.000002%, but Little Data analysis is giving them the confidence to stick with it
This approach is a classic example of Clayton Christensen’s Disruptive Innovation model in that it requires far less storage and computing power and chips away at Big Data from below with a “good enough” solution. It should also be noted that outside the US, there is a huge “bottom of the pyramid” market of over 200 million businesses that have 100 or fewer pieces of data.

Our firm currently has two investments in this space, Unidata and FewPoints, and are on the verge of a big announcement on a third that I can’t speak about at this time.