Identity resolution for connected TV: Reaching real people in a fragmented streaming world

While CTV has largely surpassed linear TV as the preferred method for reaching a brand’s best customers, CTV has more screens, more apps and more of a fragmentation problem. That gap is exactly what identity resolution solves—and it's the single most important concept to understand if you're serious about CTV advertising.

In this guide, we’ll break down everything you need to know about identity resolution as it relates to CTV. Let’s dive in.

The CTV identity problem: Why streaming broke targeting

Streaming overtook traditional broadcast and cable in total share of TV viewing time during 2025, and that lead has held through early 2026—Nielsen's The Gauge has shown streaming consistently capturing roughly 47% of total TV viewing time, ahead of broadcast and cable individually.

That's the good news for CTV as a channel. Here's the problem it creates for advertisers: the average U.S. household subscribes to multiple streaming services simultaneously. Each one runs its own ad stack, its own measurement and often its own walled-off identity system. Without identity resolution tying those exposures together, the same household can see the same ad anywhere from 8 to 15 times across different services—and the advertiser has no way to know it's happening, let alone correct for it.

This is the core problem identity resolution solves.

What identity resolution actually means in CTV advertising

Identity resolution is the process of connecting disparate identifiers (device IDs, IP addresses, hashed emails, household graphs, registered logins) into a single, persistent view of a person or household.

It's worth being precise about a distinction that gets blurred constantly: an identity graph is the underlying linked-data asset—the actual record of which identifiers belong together. Identity resolution is the ongoing process of matching new signals into that graph. The graph is the map; resolution is the act of drawing new lines onto it.

Five identifier types show up regularly in CTV bid streams, and each comes with real tradeoffs:

• IP address: Household-level, but shared across a home and increasingly anonymized by carriers and devices, making it a weaker signal on its own than it used to be.
• Device IDs: IDFA/AAID-style identifiers on streaming sticks, plus gaming-console-specific IDs.
• Hashed emails: Increasingly the most durable signal in a post-cookie landscape, since they don't depend on browser or device behavior.
• Registered IDs: Logged-in identifiers from services like Hulu, Netflix or Amazon profiles.
• Identity graph IDs: Proprietary, persistent IDs from vendors like Epsilon COREid, or The Trade Desk's UID2.0.

No single signal here is enough on its own. That's exactly why resolution exists—it's the discipline of combining multiple imperfect signals into something more reliable than any one of them alone.

Deterministic vs. probabilistic matching—and why most real systems use both

Deterministic matching relies on known, verifiable links, like a hashed email matched directly to a streaming login, for example. It's the highest-accuracy approach, but it has a ceiling: you can only match what you can verify, which limits scale.

Probabilistic matching works differently. It infers connections statistically from co-occurring signals—IP address, time of day, device type and content-viewing patterns, all considered together. It scales further than deterministic matching can, but with an accuracy tradeoff.

In practice, most modern identity systems blend both—deterministic anchors extended with probabilistic reach where verified data runs out. The decision framework is fairly simple: deterministic-only is enough for small-scale, high-trust use cases, like one-to-one messaging to a known customer list. Blended approaches become necessary for CTV reach campaigns, where the scale requirement makes deterministic-only matching impractical on its own.

Household vs. individual identity—a strategic choice with real tradeoffs

TV has always been a household medium—one screen, multiple viewers, one set of eyeballs counted together. CTV inherits that household framing, but it also adds something linear never had: the ability to identify individuals within that household through app-level logins.

Household identity treats every viewer in a home as a single audience unit. This works well for brand-suitability concerns, for frequency capping on shared screens, and for family-relevant categories like QSR, auto, and retail, where the purchase decision often involves the whole household anyway.

Individual identity differentiates within the household by app login, which matters for streaming-service-level targeting, genuine personalization or B2B targeting where one specific decision-maker in the house is the actual audience.

There's no universal right answer—it depends on the campaign. The best identity platforms are built to support both, and which one you lean on should follow from the campaign's actual goal rather than from whichever signal happens to be easiest to get. It's also worth noting that CTV inventory itself varies in which identifier gets exposed. Some publishers send IP only, some send logged-in IDs, and that variability is part of what makes a flexible identity approach so important.

Walled gardens and the identity resolution challenge

The biggest CTV performance streamers—Netflix, Amazon Prime Video, Disney+, Hulu, YouTube TV, Max—each maintain their own closed identity systems. You generally cannot bring your own identifier into Netflix's ad ecosystem and target against it directly. Instead, you bring your audience to Netflix, and they handle the matching on their side, inside their own walls.

Clean rooms are the industry's emerging answer to this problem. They’re privacy-safe environments where an advertiser's data and a publisher's data can be matched without either side directly exposing their raw data to the other. Disney's Real-Time Ad Exchange, Amazon Marketing Cloud and NBCUniversal's One Platform Total Audience are current examples of walled gardens building out this kind of collaborative infrastructure.

Epsilon approaches this through Epsilon Clean Room, which is built specifically to bridge identity, data and activation across these kinds of partner environments, which enables advertisers to activate inside walled gardens without losing the identity consistency that makes cross-platform measurement possible in the first place.

Privacy, consent and the post-cookie identity landscape

The privacy landscape that CTV identity operates in keeps getting more complex. State-level privacy laws like CCPA/CPRA in California, plus a growing list of other state laws, continue to expand the compliance requirements identity providers have to navigate.

On the cookie side, the long-anticipated story took an unexpected turn: Google ultimately abandoned its plan to forcibly deprecate third-party cookies in Chrome. As of 2026, Chrome retains third-party cookies under a user-choice model rather than phasing them out by default—a meaningful reversal from years of will-they-won’t they.

Here's why that reversal matters less for CTV than you might expect: CTV identity never depended on third-party cookies in the first place. It was built from day one on device IDs, household graphs, hashed emails and first-party data matching—none of which a browser-level cookie policy touches. That's not a coincidence; it's a structural advantage. The durable stack going forward is hashed-email matching, first-party data and clean room activation, which is the same stack CTV has relied on all along…regardless of what Chrome ultimately decided to do.

What to look for in a CTV identity partner

If you're evaluating identity providers, six criteria are worth running through as a checklist:

1. Scale of the identity graph. How many U.S. households and individuals does the graph actually cover?
2. Match accuracy. Is it third-party validated, or just self-reported by the vendor?
3. Privacy posture. What does their DPA (data processing agreement) look like, how is state-law compliance handled, and what opt-out mechanisms exist?
4. Walled-garden partnerships. Does the provider have real clean-room integrations and streamer alignments, or just aspirational ones?
5. Cross-channel persistence. Does the identifier hold up across linear, CTV, digital, and in-store, or does it break down outside one channel?
6. Anchoring methodology. What's the identifier actually anchored to? Physical-address-anchored identity tends to be more durable over time than identity anchored purely to email or IP, since addresses change far less often.

Epsilon's approach holds up well against each of these. Our CORE ID is anchored in physical addresses rather than relying solely on emails or IPs, spans both households and individuals, and is integrated across Epsilon CTV activations as well as major walled-garden environments—which is precisely the combination this checklist is built to surface.

How COREid solves CTV identity resolution

CORE ID's physical-address anchoring is the foundation of its durability: it doesn't depend on a browser policy, a device refresh or a login persisting. It spans households and individuals, which means it supports both of the targeting strategies covered earlier in this guide rather than forcing a choice between them. And it's integrated directly across Epsilon's CTV activations and major streaming partnerships, which is what makes the walled-garden challenge described above tractable in practice rather than theoretical.

This approach was recognized in the IDC MarketScape: Worldwide Connected TV Advertising Platforms 2025 Vendor Assessment, where Epsilon was named a Leader—with the report specifically citing Epsilon's identity-based targeting and COREid's role in reducing wasted ad spend through more precise, deduplicated reach.

If identity is the piece you're trying to get right, explore Epsilon's identity capabilities or Epsilon's CTV advertising platform.

FAQ

What is identity resolution in CTV?

The process of connecting disparate identifiers (IPs, device IDs, hashed emails, login IDs) into a single, persistent view of a household or individual, so an advertiser can target, frequency-cap and measure consistently across streaming services.

How is identity resolution different from an identity graph?

An identity graph is the underlying data asset—a linked record of identifiers tied to a person or household. Identity resolution is the matching process that connects new signals into that graph.

Why do you need identity resolution for CTV when you didn't need it for linear TV?

Linear was bought by demographic proxy via panels. CTV is bought against actual households and individuals via real bid-stream signals. Without identity resolution, you can't deduplicate frequency, retarget consistently, or measure outcomes with any confidence.

Is identity resolution privacy-safe?

When implemented with proper consent management, hashed identifiers, and clean rooms, yes. Modern identity resolution is generally more privacy-safe than the cookie-based identity systems it's replacing, since it depends on consented, governed data rather than passive browser tracking.

What's the difference between deterministic and probabilistic identity resolution?

Deterministic matches are based on known, verifiable links, like a hashed email matched to a login, for instance. Probabilistic matches are statistical inferences from co-occurring signals. Most production systems use both.

Can you do identity resolution inside walled gardens like Netflix or Disney?

Not directly—walled gardens don't expose their internal identifiers to outside partners. Instead, you activate identity inside walled gardens via clean rooms, like Disney's Real-Time Ad Exchange or Amazon Marketing Cloud, by matching your audience against theirs in a privacy-safe environment.