Wednesday, May 12, 2010

I have a dream ...

I have a dream. My dream is that one day those overpaid nincompoops who run many of our companies and organisations wake up to the importance of data, and start working with it accordingly.

If you're not persuaded of the importance of data, try imagining your organisation functioning without data (or its cousin, information, which is usually rooted in data) and see how far you get. No e-mails, no internet, no customer orders, no invoices. No telephone calls, no meetings, no discussions with colleagues, not even to discuss the weather, unless you're one of the very few organisations which is not affected by the weather (really, you'd be surprised).

How long would that situation be able to last? Minutes?

Why can't people understand the importance of data and its quality? Why don't we treat it in the same way that we treat other parts of our business? The very idea of an airline only maintaining its fleet when something went wrong with it would horrify all of us, but that's what we do with data. Few of us do not realise how preventing tooth decay not only saves us costly treatment and potentially a great deal of pain, but leaves us with far better teeth than any dentists ministrations could produce on badly maintained teeth. (Read Jim Harris' blog post on that topic here.)

So why do we wait until the CEO is told that $ 1 billion PROFIT was made instead of the actual $1 billion LOSS, with the resultant chaos, before we take data seriously? Clearly, unmaintained airlines falling from the sky make a greater immediate impact than data quality wrecks, but the results can be equally pernicious. Why must so many people waste so many hours trying to prove return on investment (ROI), when ANY and ALL data quality improvements are beneficial - I am yet to be persuaded that there is no return on any investment (in one form or another) on every improvement of data quality. Sadly, most businesses make money DESPITE their data quality, not because of it. (See Henrik Liliendahl Sørensen's post showing how simple it can be to show ROI here).



I have a dream of a revolution in data quality, where resources and focus are built into the prevention of data quality problems, rather than on trying to resolve them only when their detrimental effect becomes obvious; where as much control is put into data as is put into production, maintenance, finance, human resources and other aspects of organisations.

I have a dream. How long must it remain a dream?

3 comments:

Phil Simon said...

I have hope that others are coming around to DQ. Change takes time, though. It's frustrating that others don't see the light. I suppose that we are enlightened!

Robert said...

Proving ROI may be an unwelcome task for gaining approval for data quality projects, but it is a time honored way of allocating corporate capital. Most corporations have limited resources and must use capital budgeting to allocate those resources. Clearly, data quality can provide excellent ROI, but it is incumbent on the project's leaders to show management the benefits of data quality.

I think that your dream will come true as businesses see how other companies such as Harrah's and Capital One have used analytics to become more competitive. This use of BI will create a greater need for data quality.

Graham Rhind said...

Robert: That something is always done, and always has been done, is not a good reason for continuing to do it. Corporations may have limited resources, but that will not be resolved by holding and using bad data - in fact, it will just prolong that state of affairs.

I am pretty sure that the leader of the maintenance department of an airline does not have to prove to the CEO why they are needed - it is self-evident what damage would be caused if aeroplanes were not maintained. It is self-evident, to me at least, how better data quality will contribute to an organisation's success. In 20 years I've never come across an improvement in data quality which has not paid for itself many, many times over. So why do we STILL have to prove it? I'm afraid I put this down to bad management, and there's plenty of that around.

Note: I am talking about data quality improvement, not projects to improve data quality. Some projects are ineffective, over-priced and wasteful, and you don't necessarily get your money's worth. But it is really very simple to get 90% of your data quality issues resolved quickly and cheaply, and THAT'S worth every penny.