• by Cory Treffiletti , Featured Contributor, December 2, 2015

Do you know the difference between probabilistic and deterministic audience data — and how people are using these different types of data to build more effective marketing and advertising programs?  These two words have been thrown around quite a bit lately — and since most people in advertising hate to use words with this many syllables, they must be important. So here goes:

Probabilistic gets its root from the word “probability,” or the likelihood that something is true. In the world of data targeting, you can say a particular device is “probably” audience X, with a certain degree of confidence going as high as 99%.

Deterministic means the data can be sure it represents a specific segment of the audience. In short, you can “determine” with 100% confidence that a device is audience X.  It’s a subtle difference between the two, but an important one.

What’s interesting is, most marketers don’t dive into the depths of truly understanding the inherent accuracy of each.  Truth be told, there is actually very little deterministic data currently being used in the industry — just differing degrees of confidence.  To scale, you need both probabilistic and deterministic data to get to a dependable degree of confidence while also protecting consumer privacy.  

If you work with a partner who professes one or the other, you might be trading accuracy for scale.  If you enable both, you have to be able to match them — and matching is no easy consideration, especially when you factor in the variety of devices, channels and ID methodologies that all represent “people.”

This is why you also have to look at the additional dimensions of matching that include source type, device type, anonymous user data, known users, households, individuals and more.  Finding the right partner to match all these different methodologies is difficult and requires a skill set that has not been traditionally available in marketing.  Conversely, while this skill set does exists in analytics or IT, these folks traditionally don't have any experience in marketing, so the disconnect continues.  

Marketing teams are enamored with data and the quantitative opportunities it provides, but they need to make sure they have the right partners and the people to understand and activate it.  

I’ve recently seen a very effective model where traditionally disparate groups are working together.  In one company, both an IT employee and an analytics or data science person were embedded in marketing. Those employees had a direct line of reporting up through the CMO, and a dotted line to the head of IT or the head of analytics, respectively. 

This embedded structure creates more inherent communication between these three teams, fostering a better understanding of how to use different kinds of data and platforms.  It also enables the teams to know the right questions to ask when identifying partners to work with, ensuring those partnerships will be successful.  

This “embedded” model is one that works in other lines of business, and it seems to be gaining traction in the area of data-driven marketing.  At a recent conference where I chatted with many CMOs from top 200 global brands, they all agreed this model was worth pursuing; some had already been thinking about and testing this same path.

Much like the rest of the world, it’s about having mutual respect for what others do and seeing how we can work together. 

Most company departments are starting to gain more and more respect for marketing because we are open to trying new things l and utilizing different types of data, with the results being proof of measurable revenue impact.  

Are you thinking this way for your organization?  Share your stories on the Spin Board!