Last July, I had the honor of being the Master of Ceremonies for the French Dreamforce Event. It was part of the Big Data Discussions with some of the greatest French minds in that field, such as Romain Niccoli, co-founder of Criteo, Florian Douetteau, founder of the French Startup, Dataiku, Romain Lacombe, who is part of the Open Data Initiative for the French Gov.
During these conversations, another speaker, Alex Dayon, who is the President of Salesforce.com,shared a case study involving GE jet engines. He explained that on average a jet engine produces 2 terabytes of data a day. The data comes from hundreds of both physical and virtual sensors. The question that comes to mind is, how much of a value are these 2 terabytes of data worth? This sound more of a google interview question, but it’s also a very strategic one that every CIO should be able to answer.
The Value of 1Tb of Ebooks
The average size of a book is roughly 0.3 Megabytes. Now looking at ebooks in pure data form, we see that the total ebook market in 2012 was worth $3.34b for 457 millions e-books sold. Therefore, we can assume that the average price of an ebook is around $7.00. But most importantly, we can deduce that 1Tb of ebook data is worth approximately $25.5 Million a Terabyte.
The Value of 1Tb of Social network
Facebook Revenue for the fourth quarter of 2013 was $2.59 billion, knowing means Facebook is dealing with 600 Terabytes of data per day In that case we can conclude that one terabyte of Facebook data is roughly worth $47,305 a Terabyte.
The Value of 1Tb of Jet engine’s Data (Estimate)
This one is really tough to measure, as I don’t have enough information currently. But we could try and estimate the value of the GE jet engine data for maintenance. We know that a GE jet engine produces 2 terabytes a day. Roughly estimated, the sensors data will lead to a much faster and better diagnostic maintenance. Assuming the new engine can reduce that cost by 20 percent, following a good 80/20 rule of thumb, and the cost per year of maintenance is roughly $1M per engine (source IATA) that means a plane flying for 330 days a year would have 660 Terabytes of data worth $200k of the potential value in that context. So, 1 terabyte of Jet engine data is worth roughly $303.
The rise of huge data make sense, as the cost to store large amounts is now inferior to the value held by this data. It’s similar to the mining concept where you will not dig if it costs more than the minerals that you will extract. Since a terabyte of storage is around $60, any terabyte of data that can be worth more than that should be stored.
Each terabyte of data is different from one another, for instance, 500 pages in a glamour magazine does not have the same value as 500 pages of a dictionary, even though the glamour magazine would be worth more.
Return on Data
Knowing the potential value of data is like knowing the Total Addressable Market (TAM). These are the maximum theoretical values you will most likely never achieve. But you should be able to aim for a reasonable “Return on Data.” here defined as the Ratio between the potential value of 1Terabyte of data and the effective value extracted out of it.
The formula : Return on Data = (Total Revenue real or expected from specific Data) / (Size of Data in Terabyte)
Each terabyte of data is different, and in the race of collecting more and more data, we should also focus on extracting more and more value out of it. By measuring the Return on Data, CIOs will be able to benchmark not only the yield and quality of their data but also planning aggressive goals for their teams… And it s fun to do ☺