Deterministic Data: Powerful Tool For Predicting User Behavior

In a crowded digital marketing world, every company is looking for an advantage. Sophisticated marketers and the agencies that support them are beginning to look deeply at using deterministic data—that is, data collected firsthand, aggregated from sources like subscriptions, opt-ins, and onsite user behaviors. We can also look to external sources of first-party data like Facebook and Google to help extend a brand’s reach to its highest-value customers.

However, deterministic data is more limited in scale than its cousin, probabilistic data (where we tease out the probability of a certain audience type responding in a certain way, based on what other users with similar characteristics have historically done). It’s also really challenging to find portable sources of deterministic data outside of the walled gardens of Facebook and Google.

That said, there is no more powerful data available to online marketers today than deterministic data. Analyzed and then activated, it can provide the backbone for a well-honed marketing strategy able to granularly target in-market consumers while also measuring their interactions with a product or brand across devices.

Here are some of the best practices I recommend:

Get the most out of your data. Your data will always be the most valuable when it’s collected by you. This data reflects the exact behavior of your consumers and the way in which they are interacting with your brand. It's the best predictor of future behavior you have, so it's worth your time and effort to invest in a data management platform (DMP) to harvest and activate first-party data.

Also, look for ways to extend the value of your own data with look-alike modeling.  Compare it against qualified deterministic data sets to help remove the ambiguity around where your next ad dollar should go.

Respect the limitations of deterministic data. As powerful as a properly deployed DMP is, it is only as powerful as the size of the data set you have available for analysis. That begs the chicken-and-egg question: you want more user data to analyze, but in order to get it, you have to attract more users to your site. Emphasize the quality of your top-of-the-funnel marketing efforts to attract qualified in-market audiences and give them multiple opportunities to interact with your Web property so that you can harvest that user data and build out your data sets.

Use deterministic signals to get out from behind the walled gardens. Facebook and Google provide amazing access to deterministic data, but the catch is that you can’t leave Google or Facebook. This paradigm limits scale and inhibits marketing campaigns by relying too much on either company.

There are other ways to achieve the same level of granular targeting. Let’s say you’re selling a product targeted to commuters. Knowing who has the NextBus or Metro North App installed on their phone would be a great way to qualify a consumer as in-market for that product. App ownership data can provide unprecedented access to deterministically validated user action or attribute and extrapolate a qualifying behavior.

The best part is that the data set is portable and can be leveraged to inform buying decisions across devices and inventory sources (imagine being able to target a TV ad to a consumer based on an app they owned).

1 comment about "Deterministic Data: Powerful Tool For Predicting User Behavior".
Check to receive email when comments are posted.
  1. Joseph Dressler from LiveIntent, September 28, 2016 at 9:26 p.m.

    Last week, we were at Path to Purchase Expo in Chicago where I had the pleasure of leading a panel with AOL and Pandora, all talking about People Based Marketing.  The exciting part was talking about 3 large companies who use deterministic data at scale and even more so in an open environment.  

    We touched on multiple points including being able to share and pass data back to the marketers to help make them smarter.  Check any of those companies out for other solutions.

Next story loading loading..