With Valentine’s Day fast approaching, our inboxes have been inundated with guides revealing hot tips on how to best spend the holiday – with or without a significant other. The recipe for a “successful” Valentine’s Day date is well-documented – we’re told it consists of some combination of flowers, fine dining, entertainment, and sweet treats. Like advertising, we know that when campaigns (or date activities) are customized to personal preferences, the result is an even more memorable and impactful experience.

At Dstillery, we use data science to analyze digital and physical signals, helping you find your next audience. Can we also use these signals to understand your date night preferences and help you find your perfect match? Take the quiz below to find out.

Coffee or a Cocktail? What Your Date Preferences Tell Us About Your Perfect Match
Click the image below to enlarge.


Here’s How It Works:

Start at the top, choosing between two possible date activities. The first choice, deciding between a coffee or a drink, represents the lowest level of intensity (a first date). From there, follow the branch that corresponds with your answer to the next choice of date activities. As you navigate further down the tree, these activities begin to heat up! Once you have made your final choice, a description of your recommended match awaits you.

Try for yourself and click on the recommendation that corresponds with your last answer!

Your Perfect Match:

To activate any of these audiences, contact your account manager or visit Audience Studio.

Activate Audiences

The Science Behind The Data

We took it upon ourselves to design a simple tool to help reduce the complexity of courtship with a Valentine’s Day decision tree. A decision tree is a flowchart-like structure often used in machine learning to help visualize how predictive models make decisions. This is something we are all too familiar with at Dstillery, as we deploy thousands of models daily that are designed to predict whether or not someone will engage with an advertisement. For the task of matchmaking, we borrowed this tool to help visualize how to go about identifying your ideal partner. The intuition behind this is simple: based on your date activity preferences, we recommend the type of person that makes for the best match.

With a database of over 300 million web-based user profiles, we have a broad view into people’s behavior. Each of the date activities listed in the decision tree represents a behavioral segment we have in our system (e.g. Coffee = “Coffee Lovers”). Users can fall into multiple behavioral segments based on their web browsing behavior. Each of the sixteen possible match profiles consists of users who fell into each of the segments that correspond to a date activity. For instance, if you answered Coffee, Movie, Aquarium, Fishing, we reference users who fell into each of those behavioral segments and look at other segments they are likely to fit into.