Seedtag and IRIS.TV Bring Program-Level AI Targeting to Connected TV
Seedtag and IRIS.TV have formed a partnership to expand privacy-first neuro-contextual targeting in connected TV (CTV). The integration brings IRIS.TV’s program-level content signals into Seedtag’s AI engine, Liz, to improve contextual precision, scale, and transparency for advertisers.
The collaboration focuses on strengthening program- and scene-level targeting in CTV without relying on identity-based signals. By combining enriched content metadata with AI-driven contextual analysis, the companies aim to improve performance while maintaining privacy compliance.
Daniel Church, Head of CTV at Seedtag, outlined how the partnership enhances contextual decisioning across fragmented streaming inventory.
How does this partnership help media buyers know which programs their local ads are actually running in, not just which apps or platforms?
Contextual at scale is hard for local TV because small geos already constrain reach, and traditional contextual filters by title or genre, which can collapse scale fast. Seedtag solves this by maximizing what we know about each show and movie, using our neuro-contextual intelligence to understand themes, tone, and suitability beyond basic metadata. That richer signal lets buyers set meaningful contextual guardrails without over-restricting the available supply. Our AI, Liz, aggregates and expands across many programs and films that share the right contextual patterns, so campaigns retain scale while still aligning to objectives.
RMNs are increasingly extending into CTV. How can program-level neuro-contextual targeting help retailers align streaming ads with shopping moments or category intent without using shopper data?
Even without shopper data, Neuro-Contextual has a lot to work with because we can infer patterns and intent from open-web signals, viewership behaviors, and scene-level themes. Liz converts those signals into intent scores that are built directly into decisioning, so we can place ads in the moments most likely to drive understanding and action. Instead of targeting people based on who they are, we align to what they are watching and what they are leaning into, using program and scene understanding to stay relevant. The outcome is higher-quality contextual delivery that improves product comprehension and performance, while staying privacy-forward and scalable.
What does neuro-contextual targeting add on top of geo, daypart, and genre signals, and when does it meaningfully outperform them?
Neuro-Contextual adds a deeper understanding of how audiences respond to contextual signals, which matters for local advertisers where geo is non-negotiable. In CTV, relying on daypart and genre often limits scale, with inconsistent performance lift because “genre” is broad and defined differently by each media owner. As a result, the same label can represent very different content depending on the supply partner, making planning and optimization inefficient. Neuro-Contextual normalizes those inputs across supply and adds richer signals such as themes, tone, intensity, suitability, so targeting decisions are based on consistent meaning rather than inconsistent tags. This approach preserves geo targeting, retains more scale than strict genre/daypart, and enables context-driven decisions with higher-fidelity data.
The promise of this partnership is expanded scale through enriched metadata. For a multi-location or regional retailer, what does that look like in practice?
Enriched metadata expands scale and reach while improving relevance because we can qualify more inventory with higher confidence than title or genre alone. Those added signals provide a granular understanding of not just what the content is, but how different audiences engage with it and how it tends to influence attention, sentiment, and propensity to act. Liz aligns those signals to campaign goals, then surfaces the programs and moments most likely to deliver against those goals rather than defaulting to the most popular placements. While performance varies by objective, results consistently improve when brand messaging is matched to the right program contexts, making the ad feel relevant rather than interruptive.
Transparency remains a major concern in local CTV buys. What new visibility does IRIS_ID unlock for advertisers trying to understand delivery, adjacency, and performance across fragmented streaming inventory?
The IRIS ID is a strong identifier that standardizes content metadata across supply and improves consistency in how programs are described and reported. It also enables activation against content metadata in environments where the underlying metadata cannot be passed through directly, by providing a common reference layer. IRIS is one input within a broader set of content metadata sources that Liz uses to build a deeper understanding of what is actually on screen. That combined graph expands usable inventory while preserving contextual precision, with decisions informed by richer, normalized cues rather than inconsistent publisher labels.
With retail media and local broadcasters pushing privacy-first strategies, how does this approach maintain performance without leaning on IDs, household graphs, or deterministic data?
Privacy has often been treated as a tradeoff that weakens performance by limiting what data can be used or shared. In practice, buyers do not need to choose privacy or performance because the right signal can come from the content and viewing environment rather than identity. Traditional contextual in CTV is usually broad genre targeting, and when you push it more granular you often lose scale and risk under-delivery. With neuro-contextual, genre is just one input in a much larger dataset that captures what the program is, what it is about, and how different audiences respond to it, including how those responses shift over time. By combining multiple content and behavioral datasets, we can drive outcomes while staying privacy-forward and maintaining scalable delivery.
Do you expect neuro-contextual targeting to become a standard layer in local TV and retail media buying?
I expect it to become a standard layer in many campaign types because it enables privacy-forward targeting while addressing core CTV inefficiencies. Today, the household’s IP address is the dominant CTV identifier and it rotates frequently, and even hashed email targeting maps to the account owner, not necessarily who is actually watching. That mismatch creates wasted impressions and weak relevance, especially for outcomes-focused campaigns. Neuro-Contextual bypasses those ID limitations by optimizing to the viewing environment itself, using a rich dataset of program, scene, and behavioral signals to make smarter decisions. The result is better on-target delivery and improved outcomes –without relying on brittle household identifiers.
This Seedtag partnership with IRIS.TV integrates program-level metadata and IRIS_ID into its AI engine, Liz, enabling scalable, privacy-first neuro-contextual targeting in CTV. By optimizing against content signals rather than identity data, the approach aims to improve transparency, relevance, and campaign performance across fragmented streaming inventory.
