Horizon Media EVP & Head of Platform Partnerships John Koenigsberg’s Exclusive Interview with MarTech Pulse on Predictive AI

Can you share your career journey from leading platform partnerships at Horizon Media to championing predictive AI like ZeroToOne in HorizonOS, and what pivotal research or client challenges inspired your focus on behavioral prediction?

My path at Horizon started with founding big, Horizon’s growth agency for early-stage brands. Building go-to-market strategies alongside founders reframed what’s possible when technology is a core growth lever for the business. That work pulled us into early AI-native use cases and shaped how I think about platforms and creating integrated systems.

Leading platform partnerships for Horizon further exposed how much value gets lost in fragmented ecosystems and reactive optimization. Ultimately our north star is client growth. The opportunity we see with behavioral prediction is connecting new sources of intelligence—fresher signals and shorter prediction windows—into a platform that gets smarter the more you plug into it, giving clients a clearer picture of who’s likely to act and when to drive better outcomes.

How has integrating ZeroToOne.AI’s Large Behavioral Model into HorizonOS’s Blu platform shifted marketers from reactive workflow automation to proactive consumer behavior prediction, and what specific efficiency gains in visitation, conversion, and media waste reduction have clients seen in pilots?

The integration is a shift from optimizing against past conversion signals to identifying consumers predicted to act next—and constantly refreshing those predictions. The real value is what happens when we layer predictive signals on top of our existing data foundations. Combining behavioral prediction with our core audience intelligence and thoughtful campaign architecture is the magic behind measurably better outcomes. One recent pilot built this way drove 48% greater visitation efficiency and 55% higher ROI versus baseline.

Read More: Questex VP and Market Leader Kevin Gray’s Exclusive Interview with MarTech Pulse on Streaming Industry

Drawing from your experience leading platform partnerships at Horizon Media, including recent Newton AI-Snowflake integrations, what criteria do you use to select and prioritize AI partners that deliver interoperable, real-time predictive intelligence for scalable client workflows?

Measurable client growth is the focus, so we especially evaluate partners on whether their capabilities compound across accounts and workflows, not just perform well in a narrow context. In practice that depends on genuine technological differentiation, clean integration into open ecosystems, and repeatable lift driven by a scalable mechanism. We look closely at operational readiness; can we activate this in market, not just discuss it? And we prioritize deep collaborators with shared motivation to problem-solve together.  Finally, we believe that our best partnerships don’t just solve today’s problem, but expand what we can offer clients tomorrow.

HorizonOS Labs tests emerging tech through hypothesis-driven pilots what challenges in predictive audience targeting are brands bringing to labs, how does ZeroToOne.AI address them, and what ROI metrics prove scalability across QSR, retail, and CPG categories?

The brands coming to Labs share a hunger to win and get sharper—more relevant audiences, tighter precision, ultimately pushing the envelope on performance. They want a technology-enabled edge over competitors. ZeroToOne helps us deliver on that by refreshing intent signals rapidly with short prediction windows, particularly valuable in QSR and retail where purchase cycles are tight.  Scalability metrics include incremental visitation lift, cost-per-visit reduction, and conversion efficiency versus holdouts. Additionally, a single pilot is directional. Repeatable lift across industries is what demonstrates enterprise viability.

ZeroToOne.AI claims 85%+ accuracy in anticipating real-world actions via its Large Behavioral Model how do martech professionals validate and benchmark such predictive models against traditional analytics, and what research trends are you observing in behavioral data science?

Validation always starts with controlled experimentation, especially holdout groups, and incremental lift measurement. Any accuracy claim needs to prove out in live media against confident baselines.  As far as the research trend I’m watching most closely: the shift from observing and predicting behavior at a distance to actively shaping it in real time. As AI surfaces—search, brand agents, conversational interfaces—become places where consumers actually make decisions, behavioral science moves from interception to more dynamic interaction. That changes what prediction even means. The models that matter next won’t just tell you who’s likely to act, they may inform experiences that influence the action itself.

Read More: ActiveCampaign VP of Product Marketing Jackie Palmer’s Exclusive Interview with MarTech Pulse on Autonomous Marketing

What 2026 KPIs like real-time conversion uplift should martech leaders track for predictive AI partnerships like ZeroToOne-HorizonOS, and how can they build guardrails and self-calibration to ensure sustainable ROI in autonomous growth systems?

The KPIs themselves are only as valuable as the measurement discipline behind them. For the use cases we’ve been circling, incremental lift, cost-per-visit efficiency, sales impact, and waste reduction are all incredibly useful, but only if you’re running controlled tests and ongoing recalibrating against ground truth.

On guardrails, we think a great deal about the role of human-in-the-loop in the systems and how it must be more like a dial calibrated by risk and confidence vs. an on-off switch. Early in a partnership, you may want structured workflows with tight human oversight at every decision point. As the system earns trust through validated results, you selectively open up—letting predictive systems act autonomously where they’ve proven reliable while keeping human authority over high-stakes decisions like budget thresholds and other forms of triggered escalation. We believe the most robust “autonomous” systems will be those that pair speed with intentional human governance.

For martech readers adopting predictive platforms in 2026, what upfront research on behavioral signals and ecosystem interoperability do you recommend, based on HorizonOS pilots, to shift successfully from historical analytics to forward-looking decisioning?

First, assess whether signals refresh at a cadence that matches actual consumer behavior. Second, predictive intelligence delivers the most value when it reaches the orchestration layer (where bidding, activation, and creative decisions actually happen) vs. stopping at a reporting dashboard. Emerging AI protocols are actively dissolving the integration tax that historically kept systems siloed, making predictive signals portable across partners, channels, and extended use cases for the first time. Third, build out shared measurement frameworks that scaffold plans with clarity and accountability.

Thank you, John Koenigsberg,  for taking the time to share your insights with us.

Write to us [⁠wasim.a@demandmediaagency.com] to learn more about our exclusive editorial packages and programmes.