Several weeks ago my company's technology community came together for an EMC Tech Summit in Cambridge, Massachusetts (the Twitter feed for the summit can be found here).
One of the areas we discussed was the Data Value research being conducted in partnership with Dr. Jim Short at the San Diego Supercomputer Center at UC San Diego. Jim has been conducting related research for several years (see some of his early work from 2013). He was able to attend the Tech Summit and share some of his early findings with the broader EMC technology community.
I was taking notes and plan on sharing my thoughts on his research over the next few weeks. In this post I'd like to expand on one of the more interesting ideas he presented: ephemeral data value.
When I first became introduced to the concept of data value/valuation I thought about it in terms of what Dr. Short calls a "refinery model". Corporations are ingesting and curating massive data sets and some are beginning to treat them like a portfolio of assets. This portfolio of data assets requires a certain resource investment to maintain and curate. In addition, these companies make investments in data science teams and IT sandboxes to refine the data in an attempt to generate additional business value.
In other words these corporations leverage data to gain insight and increase revenue/cut costs. Data sets have economic or business value in this regard.
Dr. Short confirmed this view during his talk. My view as a technologist is consistent with the market view of this form of data set valuation.
But is there a market for valuation of data that has a shelf life of 10 seconds or less?
If you don't have any teenagers living at home it is possible that you aren't aware of the Snapchat app. This app allows users to share photos and videos that disappear within 1-10 seconds. The growth in popularity of this platform since September 2011 is remarkable:
Over 100 million monthly users
Roughly 70% of the users are women
Over 400 million snaps per day
71% of users are under 25 years old
Is it possible to assign economic or business value to ephemeral data that disappears within seconds of its creation? Is there a business model in here somewhere which is markedly different from the "refinery" model for data mentioned above?
According to Dr. Short the answer is a resounding "yes". An Adweek report from January of this year speculates on the daily fee that Snapchat may be charging for ads:
Snapchat, a platform that reaches a coveted teenage audience, has started asking brands for $750,000-a-day commitments, sources have told Adweek on the condition they and their employers not be named.
While this number has yet to be verified by Snapchat, Business Insider augmented the report with a comparison of this value against other corporations like American Idol ($475,000) and Sunday Night football ($627,300).
What does this mean for valuation frameworks and processes?
My first reaction is that valuation approaches designed for the refinery use case may end up being irrelevant for ephemeral data.
In addition, it seems apparent that the valuation of ephemeral data may not actually involve analyzing (or even looking at) the content itself.
The ephemeral use case was one of many that Dr. Short highlighted during his talk. In future posts I will relate other principles of data value being proposed as part of his research, and when all is said and done hope to have a more complete picture of this evolving ecosystem.