Predictive AI Audiences (from ZeroToOne.AI)

Marc Fevrier

September 23, 2026 21:38

Overview

Predictive AI Audiences from ZeroToOne.AI use artificial intelligence and continuously refreshed behavioral signals to identify consumers who are most likely to take a future action.

Rather than relying solely on past visits, Predictive Audiences identify consumers actively moving toward a future behavior, such as visiting a specific brand or category of business.

This makes them ideal for prospecting, finding potential new customers, expanding reach, and engaging consumers prior to their next purchase decision.

💡 Key Takeaway: Traditional visitation audiences help you reach people based on where they've been. Predictive Audiences help you reach people based on where they're likely to go next.

How Predictive Audiences Work & How to Access Them

ZeroToOne.AI models evaluate large-scale behavioral signals to identify intent patterns. Signals analyzed include:

Location visitation: Historical physical visits and spatial patterns
Consumer interests: Category preferences and brand affinity
Transaction patterns: Purchase cycle indicators
Digital engagement: Online research and intent signals
Demographic signals: Audience profile match data

The underlying model continuously evaluates these inputs to predict propensity within a specified window. Audiences refresh daily to account for shifting consumer behavior.

Accessing Predictive Audiences in Ads Manager

Log into GroundTruth Ads Manager.
Navigate to Location & Audience when creating or editing an ad group.
Click Browse Audience Library.
Expand the Predictive Audiences category.
Search or select your desired ZeroToOne.AI predictive segment and click Apply.

Prediction Windows

Each Predictive Audience includes a defined prediction window indicating the timeframe in which the user is expected to act.

Prediction Window	Best Suited For
24–48 Hours	Immediate visitation intent
7 Days	Active shopping or consideration cycles
14–30 Days	Longer purchase or decision journeys

⚠️ Why Window Selection Matters: Shorter windows are effective for quick-turn visits (e.g., QSR or gas stations), while longer windows support considered purchases (e.g., automotive or electronics).

Audience Types & Selection Strategy

GroundTruth targeting capabilities can be categorized across three distinct time horizons:

Past Behavior

Visitation Audiences

Reach consumers based on location historical visits.

Present Profile

Demographic & Interest

Reach consumers based on attributes and affinity.

Future Intent

Predictive AI Audiences

Reach consumers based on predicted upcoming actions.

When choosing your audience, follow a specific-to-broad selection approach:

Audience Level	Selection Guidance	Example
Brand-Specific	Start here if an audience tailored specifically to your brand exists.	Likely Walgreens Visitors
Category-Specific	If brand-level targeting is unavailable, select the nearest category.	Likely Drugstore Visitors
Lifestyle / Behavioral	Use broader behavioral segments to expand overall reach.	Health & Wellness Shoppers

Campaign Setup Best Practices

When incorporating Predictive Audiences into existing strategies, set up a dedicated ad group for testing:

Recommended Campaign Structure

Ad Group 1 (Baseline): Standard Location Visitation Audience
Ad Group 2 (Test): Predictive AI Audience

Note: Keep location parameters, budget distributions, and creatives consistent across both ad groups to isolate performance differences.

Do	Don't
Align target audiences directly with campaign objectives. Isolate predictive ad groups to evaluate performance accurately. Evaluate performance using upper/mid-funnel prospecting metrics.	Stack multiple predictive segments in one ad group without testing. Choose unrelated large audiences solely for volume. Expect identical conversion rates to narrow retargeting campaigns.

Troubleshooting & FAQs

How often are Predictive Audiences refreshed?

Predictive Audiences refresh daily using updated location and digital behavioral signals to keep target profiles accurate as real-world intent changes.

Are Predictive Audiences the same as visitation audiences?

No. Visitation audiences target users based on recorded past visits. Predictive Audiences use machine learning models to identify users likely to visit in the upcoming window.

What if a specific brand audience is not available?

If a direct brand match is unavailable, select the corresponding category-level predictive audience (e.g., use Likely QSR Visitors for a regional fast-food chain).

Can Predictive Audiences be layered with location targeting?

Yes, Predictive Audiences can be restricted by geographic parameters (such as DMA, state, or radial distance). Be mindful that layering restrictive geography may impact overall audience scale.

Need Help?

For additional support and to ensure your campaign is set up for success, please reach out to your dedicated GroundTruth representative, or our Self-Serve team (selfservehelp@groundtruth.com) to be connected to a representative.