Intelligent Enterprise

Can data quality and data integration technologies help quell insurgency in Iraq? Going by a recent New York Times news item, this seems to be the case, and serves to remind us yet again – if indeed there was any need – about the kind of profound impact data integration and quality can have on the success of any initiative, business or otherwise.

U.S. and allied troops in Iraq are facing a unique – or perhaps, from a business perspective, not so unique – problem: how to uniquely identify and tag the entities (in this case, insurgents)? As explained in the article, it would be easy to identify and weed out the insurgents were they in military uniform, but by being indistinguishable from the crowd, they hide in plain sight. Even when they are taken into custody, however, the allied forces are stymied by the lack of a reliable, centralized biometric tracking and identification database.

As the authors of the article – one a former Marine officer and the other a former assistant secretary of defense – point out, individual units in the allied forces have been "forced to concoct their own identification databases using laptops, spreadsheets and poster boards."

Given that insurgents do not hang around the same neighborhood forever, and units/officers too transfer in and out of locations, this set of individualized, non-standard and disconnected "databases" (so to speak) makes it more difficult to unerringly identify a suspected insurgent, and connect the suspect to multiple instances of insurgent activities. To compound this problem, Iraqi civilians commonly carry two or three IDs with different names, and military units carry out multiple, seemingly un-coordinated census operations whereby "houses are labeled by one unit and relabeled by the next".

Disconnected user-managed data stores, lack of common standards and procedures, operating groups working independently of each other, entities with multiple identifiers, lack of real-time validation…sounds familiar? Business or military – the fundamental issues for data management remain the same: the lack of a well-defined data architecture on account of a real or perceived lack of time, and the difficulty in proving – up front – the return on investment in terms of both money and time.

The reality is that more often than not, data quality and integration is an after-the-fact realization, and there is a heavy price to pay for that. It is true that projects cannot always be held up for data quality issues, but in such cases, the important thing is to bring up potential data issues "on the radar", lay down appropriate project assumptions, and make sure that the "business" is aware of the potentially enormous impact that data quality can have on the success of the project. And, of course, strive for continuous improvement in data quality and availability.

Rajan Chandras is a consultant with a global IT consulting, systems integration and outsourcing firm, and can be reached at rchandras@gmail.com.