Blog: Dan E. Linstedt

« Business Intelligence and the IRS | Main | Averages and Outliers - Where's the REAL business BI? »

Information Valuation - Is data an Asset?

An interesting thought has been rolling around in my head for several years regarding information, and it's value (or lack there-of). I've spoken with Bill on this subject in the past, and numerous others - including a representative for the Government of Sweden.

There is (and always will be) a lot of talk about TCO of different BI systems, but these notions seem to ignore the data itself, or the measuring of value of the data sets. This entry in my blog is filled with unanswered questions, and if you are an information asset evaluator, insurance adjuster, or in the financial industry, I'd love to hear your comments on the subject.

It all started one day when I was teaching "How to manage your ever-growing data sets" at a Wilshire conference. I was discussing the nature of data in a VLDW/VLDB context, and talking about a specific case that Herb Edlestein had spoken on - the value of Data Mining to impute (or compute/fill in) missing data elements as part of a data quality solution.

I discussed this for a while in the context of a magazine marketing company (which is also Mr Edlesteins' example). It goes a little like this:

What if you're a magazine subscription company who makes money off advertisements that are sold in specific magazines, and you want to offer complimentary subscriptions (one years worth) to a few of your best customers. In your data set you have a table which contains subscribers, in order to offer subscriptions - the algorithms devised utilize the gender column as a heavy weighting factor. Among the millions of subscribers, there are two in particular that we'll focus on. One: Sam Smith, and the other: Sam Smith #2 - for which we have the addresses, and their magazine subscriptions, but do not have their gender.

To make matters worse, we don't have any other subscribers at their address. All we know is the following:
Sam Smith #1 - also subscribes to Hot-Rod magazine
Sam Smith #2 - subscribes to CIO Magazine

This isn't enough information to offer an appropriate free subscription, and currently the VALUE of the gender column is 98% of our weighting factor. Now under further mining, or statistics we find that Sam Smith #1 also subscribes to Young Bride, and Sam Smith #2 also subscribes to FHM. The certainty of "setting the gender column" is better, at least for the purposes of adding a subscription that each Smith would like.

Suppose Sam Smith #2 is part of our largest revenue producers, (Sam Smith #2 re-sells) what's the tangible value of the gender column when it's filled in? Is it the value of the revenue that the customer is bringing to the table?

If we make Sam Smith #2 angry by sending them an unwanted subscription (because we impute the wrong value for the gender) and they leave us for our competitor?

The real questions: What's the value of the gender column, and if we can establish a value, is it a weighted value based on different factors for each customer? Can we compute an average, and a max and min value for the entire set of customers? If we can compute a value for the data set, then can we insure it as an asset on the accounting books? Are there companies that insure data as an asset? Which of course raises the question: which data is an asset, and which data is now a liability and what's its value?

Valuation of data or information is a very tricky process, however it appears as if some companies and some governments must accomplish this in the generic sense. It would be an interesting study to conduct to find out who has already worked on this concept. I invite all readers to share their opinions.

Share: