Predictive audience targeting is positioned to change everything for the world of digital advertising. Where once we could only target people based on their past behaviors, predictive audience targeting allows you to target based on what people will likely do in the future.
As the paradigm shifts toward prediction and away from historical data, “predictive AI” has already started to become a marketing buzzword. “Predictive audiences” or “AI audiences” are starting to gain steam in different settings and platforms, while never actually providing meaningful predictions.
We’re here to set the record straight: Not all “predictive” solutions are actually predictive. In fact, nearly all predictive AI advertising solutions on the market today are less about actual prediction and more about repackaging lookalike audiences.
That’s where GroundTruth and ZeroToOne.AI‘s Dynamic Intent Predictive Audiences stand apart: we make actual predictions backed by a Large Behavioral Model, not just guesses at who might buy.
So what exactly does that mean? How do Dynamic Intent Predictive Audiences actually target future intent? And why does this approach outperform lookalikes?
Here are your answers.
The Lookalike Limitations
A standard lookalike audience is built based on a wide variety of online consumer data points: where they’ve browsed and bought, demographics, interests, and other behaviors. Those data points are then correlated across different consumers to create an audience that looks like your ideal buyer. The problem is clear: the underlying data used to construct these audiences is based on past behaviors, so it doesn’t catch when consumer behaviors have changed.
For example, if a member of a Coffee Aficionado lookalike audience decides one day that they’re going to cut back on caffeine and switch to tea, they may (and likely will) still appear in the lookalike audience. Why? Because most lookalikes aren’t updated frequently or, worse, are completely static, never receiving new data after the initial audience is created. So those lookalikes that refer to themselves as “predictive audiences”? They’re not getting the basic data required to make actual predictions.
In contrast, our Dynamic Intent Prediction Audiences are updated every 24 hours with new information and new consumers, capturing the moment that intent turns into action or vice versa. That means that we know when that Coffee Aficionado switches to tea. A few missed coffee shop visits or skipped bagged coffee purchases will indicate that this person is scaling back, so they’ll be removed automatically from our predictive audience. If they start drinking coffee again, they’ll automatically be added back to the audience.
This ensures your ad budget only goes to the highest-intent consumers, AND you’re the first to know when someone becomes likely to buy.
But that’s only half the story.
Targeting Future Intent
What really sets Dynamic Intent Predictive Audiences apart: they predict future action with 90% confidence. That’s something no other audience can match, Lookalike or otherwise. Every Dynamic Intent Predictive Audience provides a time window, ranging from 24 hours all the way up to 30 days, within which the described action will occur. For AI Predicted In-market Business Travelers – next 30 days, as an example, that means that within 30 days, any given member of this audience is 90% likely to engage in business travel activities. It’s that simple.
From there, all you have to do is select the audience and run your creative, same as any other. The results so far are speaking for themselves: We’ve already seen campaigns yielding 48% lower cost-per-visit and 60% increase in conversions versus baseline audiences.
While these results and capabilities represent a major departure from Lookalikes, the underlying way that we get there differs significantly as well. At the core of Dynamic Intent Prediction is ZeroToOne.AI’s Large Behavioral Model. This model leverages over 20,000 behavioral signals across 2 billion unique mobile advertising IDs to create sophisticated consumer profiles that are complete with an index on how likely they are to take a given action (likelihood to visit a coffee shop this weekend, for example). Once this index reaches a high enough threshold for a given consumer, the model automatically adds them to your audience, allowing you to serve your ads to a coffee shop visitor before they actually visit.
To sum it up, Dynamic Intent Predictive Audiences outperform Lookalikes for a few reasons.
- They are dynamic, constantly refreshing with new consumers instead of allowing data to get stale.
- They allow you to target people based on whether they will act tomorrow, this week, next week, or this month, instead of simply how they acted in the past at some point.
- They are both larger and more targeted, allowing you to layer more targeting parameters for greater specificity and scale over time.
We currently offer 40 in-market audiences across verticals as diverse as grocery, sports, outdoor gear, dining, hardware, and dozens of others, with many, MANY more on the way, so stay tuned!
Try Dynamic Intent Prediction Right Now.
The future of advertising is now available to all GroundTruth Ads Manager users. Not in a few years, months, or weeks. Right here, right now. And you can launch a campaign that leverages these audiences in minutes.
Give it a try for yourself. Create your GroundTruth Ads Manager account today, and start exploring our platform (no minimum spend required). And, if you have any questions, don’t hesitate to contact us.
