This week we explore two different themes related to data. The first is how to value big data. The second looks at approaches for quantifying individual experience within the context of gender discrimination.
Treating Big Data as an Asset
It’s generally taken for granted that data is a strategic asset. Every salesperson knows this, which is why they covet their Rolodex. It’s one thing to know that data is valuable, and it’s another to quantify that value and put it on your balance sheet. Mark van Rijmenam of Datafloq says 20% of large UK companies put big data on the balance sheet as an intangible asset. It turns out that this trend is fairly common throughout the EU as this report by the CEBR demonstrates. That’s generally not the case in the United Staes, where GAAP “prohibit[s] companies from treating data as an asset or counting money spent collecting and analyzing the data as investments instead of costs.” (WSJ) That said, the US “Bureau of Economic Analysis announced … the decision to recognise expenditures … on research and development as fixed investments in the national accounts from July 2013 onwards.” (CEBR)
Despite accounting’s conservative approach to data, it’s clear that data will become a valid accounting asset. The question is how to value it. Doug Laney of Gartner provides a framework for approaching this problem. The CEBR report also gives suggestions on how to go about assigning value. While it makes sense to treat data as an asset, the question in my mind is if the data is effectively utilized to drive revenue growth, does the data retain its value on the balance sheet or is that a form of double counting?
Either way, this trend will likely nudge business strategies further toward open methods and closed data. This is what we see in finance, where many trading methodologies are public and the real key is high quality data. This approach is echoed by Lukas Biewald, when discussing the rationale around why Google open sourced TensorFlow:
A company’s intellectual property and its competitive advantages are moving from their proprietary technology and algorithms to their proprietary data. As data becomes a more and more critical asset and algorithms less and less important, expect lots of companies to open source more and more of their algorithms.
Read about my thoughts on TensorFlow.
Diversity has been a hot (button) topic in the tech world for a while now. A recent piece by Jessica Valenti discusses how everyday slights to women are just as bad as overt sexism. What caught my attention is that “there’s no real way to prove that it’s (conscious or unconscious) discrimination” despite women knowing “through experience exactly what’s happening.” Surely there’s a survey methodology that can be used to quantify this sort of experience. The question is whether the experience is considered too subjective to quantify. There are clear dangers to presenting qualitative questions in a quantitative style, so it’s worth steering clear of these issues. My hunch is that in aggregate, a Likert scale approach would be sufficient to at least raise awareness and show in aggregate that the phenomenon is real. And if you’re working in R, be sure to leverage the survey package for analyzing the results.
If you have ideas on this, sound off in the comments.
When I was growing up the world was abuzz at the prospect of cold fusion. Unfortunately those claims were invalidated, resulting in a cold winter for fusion research. Much has changed, and now there is not one, but three fusion reactors vying for funding and glory. It’s exciting to think that clean and abundant energy could truly be in our future.
Great insight on the accounting treatment of data. Valuation today remains largely centered around a tangible cash flow and I suspect this methodology will likely be retained for valuing data.