Where everyone seems to be going wrong with information valuation - and a possible way forward

I've been endlessly fascinated with the problem of information valuation since before I came to the information industry.

I've been reviewing the literature about it because I think that if we can crack the problem, and operationalise it, most of the problems we have with funding go away.

The bulk of the literature (so far) breaks approaches into 3 categories that I'm going to paraphrase again -

1. How much you could sell it for (a market approach).

2. How much it cost to create it (a cost driven approach).

3. The value of using it (a utility based approach).

Where they all seem to go wrong, is that in every reading of them there's a underlying assumption of what economists might call "homogeneity" - which just means that they expect it to all be relatively the same - to some degree or other.

While at some level, we can always describe the information we have using categories, the problem is that the utility of almost all business information is extremely situational.

While we might be able to say "we've got 50 landcruisers" - and value them all the same way, we can't do the same thing with business information, because if we have information about 50 transactions, even if they were for exactly the same thing - the transaction information for 50 landcruisers - the utility of that data to our organisation is going to be very specific. While we might be able to sell it to someone who has a landcruiser maintenance business, the value that someone will pay for it isn't going to be particularly helpful for an internal valuation.

An internal valuation will need to focus on the possible liabilities that arise out of not being able to produce that information, the difference in valuation over the lifecycle due to downstream value, and the ability of the organisation to reproduce that information if required.

Just to illustrate what I mean, imagine if we lose the information about a landcruiser we've just sold - but not yet delivered, for a little while, the value of the information about that transaction is the value of the landcruiser that we sold - because it creates for us a liability to the value of one landcruiser. After we've delivered the landcruiser, the value might drop to the value of whatever service liability we still have (promised servicing, warranties etc.). If we lost this information, we also have to consider the cost of recreating it (if that was possible at all) so that we could follow through on the outstanding service labilities - and what the cost of those is.

The point is, that none of the approaches above capture these things.

Add to that, the fact that most of the information work that I do is with government - for whom the conversation is much more about public good, and I think we have a challenge for all of the valuation models that we have currently.

The way I'm thinking about the problem at the moment, is that it's very much like trying to value a set of keys - the value of the keys is contingent on the locks that they fit in, but they only fit in a specific lock, and while aggregating the data does provide some benefit, it's a secondary benefit - which valuation needs to encompass, but as a secondary priority.

In case this is starting to sound complicated - it is, but I'm not finding a simple way to approach it.

I'm also having a hard time working out how to implement it, and who will respond to which numbers - government organisations for instance, just aren't going to sell their information, so that options is meaningless. Cost of creation will dramatically overvalue information for the vast majority of its lifecycle. Where we might start to get somewhere though, is the value of using it - if we consider the risk we incur by not using it.

To get to this, I'm starting to lean is into some risk quantification methodologies like FAIR - Factor Analysis of Information Risk, and the risks associated with not having the information when it's needed.

The idea behind FAIR is that by collecting real world data (by going and asking people) about the frequency and actual cost of a risk event, and extrapolating from there, we arrive at numbers that let us quantify risks in dollar terms over specific periods of time.

It's a complicated way to get there - but it's the only thing I can think of at the moment that's both objective, rigorous, and has some chance of being accepted by many organisations in many industries.

What do you think?

Have you seen information valuation done well? At all?

How do you think about it?