By Michael Foster

One of the most astounding things about the modern world is just how much data we capture every day. For instance, most of the photographs taken in human history have been taken in the last five years, even though the camera was invented over a century ago. And the only reason there are so many more pictures now is that cameras are ubiquitous – it’s easy to take a quick snap of your lunch, so why not?

And it’s almost as easy to share that picture with your friends, family and strangers. And in turn, it’s easy for them to share that picture. It’s a bit harder – but not impossible – to track who shares what pictures of what foods to whom and collect those statistics in a database.

That’s where things get fuzzy.

If we could collect all of the pictures of all of the lunches in the world and put them all into one giant database, what could we do with that data? Obviously, provided we had the right technology, we could see what food everyone in the database had eaten. Could we use that data to predict what people would have for lunch tomorrow? In a week? In a year? Could we predict what they will have for dinner? Maybe we could discern other things about them, like what kinds of cars they like to drive, their political affiliations and whether they’re single, dating, or married?

Now things are getting really fuzzy. Clearly, what you had for lunch isn’t 100 percent correlated with what your significant other looks like and the relationship between the food we eat and our political affiliations is even less obvious. What about other connections – are there statistically meaningful correlations between facts we know and facts we want to know?

Welcome to the world of data-driven analysis.

This statistical approach to collecting large data sets has ballooned in popularity over the last decade, buoyed in large part by the proliferation of data thanks to the cellphone and always-connected computers. It has become technically simple to track what people do because they make records of what they do themselves. These actions provide the opportunity for data-driven analysis to uncover insights.

There are just two problems with this. Firstly, a lot of false positives show up when you analyze these massive data sets. And, secondly, identifying what data is predictive and what data is not is really, really difficult.

This complexity is why marketers need to learn the power and limitations of data, and that’s a key theme at this year’s DMA &THEN conference. Join us for “What Will Data Drive Next?: Future-Proofing Your Business.” Experts from Upfront Ventures, 3M, Tough Mudder and Annalect will discuss how data will drive new efficiencies and improve marketing results. They will also take a look at cutting-edge data platforms and services designed to determine what data sets are useful and which are not.

We’re seeing a massive shift in the way marketers use and think about data – attend DMA &THEN to learn what data-driven marketing executives have already learned, and how executives, consultants and analysts can use data to improve business outcomes and predict the future.

This article is brought to you by &THEN, DMA’s annual event. Click here to join the leaders of the marketing community and explore how data transforms marketing in New Orleans, October 8-10. Save over $300 when you register online.