Tuesday, April 28, 2009

The Matrix: The Coming Era of Database Marketing

Happy to inform my regular readers that Aqumena is progressing well.

As we at Aqumena dig into a wide variety of client situations, I encounter epiphanies that I then catalogue for further delving into.

A blog posting is to me is the most useful initiation of further investigation as it potentially co-opts many into the act.

In our marketing analytics practice, we were discussing the promise of integrating databases including operational sales data, post-but media data, data from syndicated studies like the retail and consumer panels etc when this epiphany about the Matrix gripped me.

The thought is that as soon as we accept a database, a research study or a data stream as representative of reality we are in fact integrating the available data on to the entire universe.

For example if we accept the NRS (or the IRS) as representative reality consisting of say, 500 million people, what in reality it means is that we are fusing, assuming the sample size of the study is 100,000, to 5000 individuals in the universe of 500 million people.

Subsequently if you accept a consumer panel of 10,000 household as representative of reality you are accepting that every record in the panel represents 20,000 households in your universe consisting of 200 million households (on the formulation of an average household size of 2.5 members from your 500 million individuals of age 15+ represented by the readership survey).

Marrying the picture of reality that the readership study presents with the picture of reality the consumer panel presents is an unquestionably valid activity because both are true representations of the same reality.

Based on the same principles, a U&A study of 5000 sample size or a price sensitivity study of sample size 2000, as long as they are accepted as valid representations of reality, can be married to the readership and panel data, individual by 500 million individuals.

The marrying of data (variously known as data fusion, data threading and data integration) is done through complex algorithms that dance to accommodate and integrate such that each view of reality that a data stream offers is valid within the overall picture of reality that is being produced.

The technology of data fusion is a developing technology but there can be no quarrel with the basic validity of the concept of welding together various views of reality, as represented by valid data streams, into a comprehensive view of reality that gets richer and more detailed as the data streams increase.

The culmination of this simulation of reality is the Matrix, the creation on a vast computer database a simulation of reality in all its complexity.

In previous times computer power was a severe limitation to simulation exercises.
I remember trying to simulate a cricket match when I was a student at IIT, Bombay back in 1976. The computer was an IBM 360. It took me months to do what can be done today on a low-end notebook in about 10 minutes.

Today weather modeling systems, with increasing effectiveness, model trillions of individual parameters using supercomputers and an array of sampling protocols across the world.

Modeling systems that consist at most of 7 billion parameters (assuming a world population of 7 billion) based on an array of operational and sampling based data streams is well within the affordability range of a major global consumer company.

Modeling a consumer universe of 200 million people is well within the affordability reach of a large size consumer company in India.

Marry the Matrix to the addressability that the emergence of the Internet,the mobile phone and the RFID chip represent and you can perceive the dawn of the age of micro-marketing.