CTV Geo-Targeting: A Technical Guide for Media Buyers
CTV Geo-Targeting: A Technical Guide for Media Buyers
The foundational location signal for CTV advertising — IP address geolocation — achieves 13% accuracy at the household level. That stat comes from a 2025 CIMM/Truthset study of one billion IP addresses. It is the honest starting point for understanding how CTV geo-targeting actually works.
If your programmatic CTV buy is built on IP-level household targeting, a meaningful portion of your geo-targeted impressions are landing in the wrong geography. That affects reach estimates, attribution models, and the answer you give when a client asks whether the campaign worked.
This article covers the full technical stack: how location data enters CTV systems, what targeting precision is realistic at each geographic tier, how attribution works, and what platforms are not telling you about the gaps.
What you'll learn:
- How location data actually flows from device to streaming ad server
- The four methods platforms use to map your audience to CTV households
- Why IP geolocation accuracy matters for your campaign performance
- How to connect CTV ad views to real-world conversions
- What to expect from each major CTV platform for geo-targeted campaigns
Reading time: ~17 minutes Level: Intermediate — assumes familiarity with programmatic advertising basics
Why CTV Geo-Targeting Demands Attention Now
U.S. CTV ad spend reached $23.6 billion in 2024 (IAB), with projections reaching $46.9 billion by 2028. Digital video spend surpassed linear TV in 2024 for the first time. Streaming now accounts for 44.8% of total TV viewership. Ninety percent of U.S. households used at least one connected TV device monthly in 2025, and 69% of advertisers classify CTV as a "must-buy."
The targeting complexity is also real. Audiences are fragmented across 40-plus streaming apps and multiple device ecosystems. Attribution methodology varies by DSP. And the location data powering household-level geo-targeting carries accuracy constraints that affect every campaign decision below the city level.
OTT vs. CTV: The Distinction That Changes Targeting Strategy
These two terms are used interchangeably in the industry — that's a problem, because they describe fundamentally different things.
OTT (Over-The-Top) is a delivery method: streaming over the internet without cable or satellite. It covers smartphones, tablets, laptops, and connected televisions.
CTV (Connected TV) is a device category: internet-connected TV screens, including smart TVs, Roku and Fire TV sticks, Apple TV, and gaming consoles. CTV is a subset of OTT.
The targeting implication is significant. OTT campaigns reach viewers across all screen types — devices that carry live GPS signals. CTV campaigns target only the TV screen, which cannot report GPS coordinates. CTV location is always inferred, never direct.
Mobile OTT is purpose-built for real-time proximity targeting. CTV is purpose-built for household-level brand reach. A campaign that geo-fences competitor locations on mobile and extends the message to those households' connected TVs is using each channel for what it does best.
How Location Data Gets Into CTV Platforms
Platforms describe geo-targeting as if it were straightforward; the mechanics reveal why it is not. There are four primary methods by which location data enters CTV ad systems, each with a different accuracy profile.
IP Address Geolocation: The Foundation
Every internet-connected household has an IP address. When a viewer's CTV device triggers an ad request, the device passes its IP address in the request header. The ad server queries a geolocation database, maps the IP to a geographic area, and matches against geo-targeting parameters.
Accuracy degrades sharply at finer geographic levels. Country-level accuracy is approximately 99%. State-level is 90% or better. City-level drops to 75–85% in dense urban markets and lower in rural regions. At the ZIP code and household level, the 2025 CIMM/Truthset study is definitive: IP-to-household linkage accuracy averages 13%, with data providers agreeing on those linkages only 6.4% of the time. IP-to-postal accuracy by U.S. state ranges from just 5% to 17%. Poor IP-to-household accuracy can reduce the effective value of a $1 media investment to $0.09.
The 13% figure doesn't mean the other 87% of impressions land in wrong households. It means the match is unconfirmed for 87% — the platform cannot verify those impressions landed where it predicted. That distinction matters for how you frame attribution to clients.
One compounding problem: most home internet users have dynamic IP addresses that change periodically. An IP-to-household mapping confirmed today may be stale within 30 days. Industry best practice is to build CTV campaign windows no longer than 30 days when relying on IP-level household targeting.
ACR Data: The Content Intelligence Layer
Automatic Content Recognition (ACR) is embedded in smart TVs from Roku, Samsung, LG, Vizio (via Inscape), and Samba TV. The television passively captures short snippets of displayed content and compares them against a reference library, identifying exactly what content is playing, from which channel or app, and at what time.
Advertisers use ACR data to:
- Target viewers who watched a specific linear TV program but have not yet seen the brand's CTV ad (incremental reach)
- Suppress CTV ads from households already exposed on linear (frequency management)
- Segment audiences by content genre — heavy news viewers, sports households, late-night viewers
- Track competitor ad frequency by geography (competitive intelligence)
ACR does not directly provide GPS location. Location is inferred through the IP address of the reporting device. What ACR adds is content context layered onto IP-level location inference — not improved location accuracy, but richer audience signal from a household already identified geographically.
Device Graph and Household Graph Matching
A device graph maps behavioral signals to individual devices, then clusters those devices into households. Shared IP address is the primary connective tissue; deterministic identifiers like hashed email addresses and first-party login data supplement it. Deterministic matches anchor household clusters; probabilistic signals (shared IP, geographic zone, usage schedules) extend coverage where hard identifiers are absent.
The key extension: if a mobile phone with GPS has been accurately associated with a home address, every other device sharing that IP — including the smart TV — inherits that location determination. This is how address-level precision on CTV is achieved. On average, providers match approximately 50% of household postal addresses to an associated IP. Major implementations include Viant's Household Graph, The Trade Desk's UID2, and Experian's Digital Graph.
Plat Line and GPS: Addressable Geo-Fencing Extended to CTV
The most precise method combines property plat line data with GPS signals from mobile devices. A mobile phone confirms the home address at GPS precision; that address is matched to the property plat line (the legal parcel boundary). All devices associated with that household through the device graph — including the CTV — become targetable at that specific address.
The operational approach: run a mobile geo-fence around a location of interest, capture GPS-confirmed device IDs, use the household graph to find those visitors' home CTV devices, and serve connected TV ads there. The person who walked into a competitor's dealership on Saturday sees your brand on their living room TV that evening. In campaigns we've tracked, this geo-fence-to-CTV model consistently shows stronger branded search lift than CTV-only buys against household IP lists.
Targeting Tiers: From DMA to Household Address
DMA Targeting (Designated Market Area)
There are 210 predefined DMAs across the United States, defined by Nielsen. On broadcast, you pay for every household in the DMA. On CTV, you bid on individual impressions within the DMA that meet your audience criteria — same geographic boundary, less waste.
State and City Targeting
IP-based targeting at state or city level is reliable — 90% or better accuracy. This is where CTV geo-targeting performs with confidence.
ZIP Code Targeting (ZTV)
Platforms like The Trade Desk allow advertisers to target specific ZIP codes, layered with census demographic data (ZTV — ZIP-code Television). A brand targeting women 25–49 in the Dallas-Fort Worth DMA — which spans more than 270 ZIP codes — can identify the 40 ZIP codes that over-index for that demographic and buy only those impressions. CPMs in ZTV campaigns typically run $10–$18, materially lower than broadcast equivalents.
Household IP Targeting
A match list of physical addresses is converted to associated IP addresses at an average 50% match rate — supporting hyper-targeted outreach to existing customers, competitor customers, or high-value prospect lists. The 13% household linkage accuracy constraint applies: treat these impressions as directional, not precise.
Address-Level Targeting via Geo-Fencing Extension
Using the plat line and household graph method, true address-level precision becomes achievable for CTV. This is the most expensive and operationally complex tier, typically reserved for high-value B2B prospecting, luxury retail, or automotive campaigns where individual household precision justifies the cost.
CTV Geo-Targeting vs. Mobile Geo-Fencing: A Direct Comparison
These two channels are frequently positioned as alternatives. They are better understood as complements with distinct roles.
| Dimension | Mobile Geo-Fencing | CTV Geo-Targeting |
|---|---|---|
| Location signal | Live GPS, Wi-Fi triangulation, cellular data | Inferred via IP address, household graph, plat line matching |
| Precision | Real-time, parcel-level GPS | Household-level; ZIP code to address accuracy |
| Trigger mechanism | Device physically enters or exits a geographic boundary | IP address matched to geographic polygon at bid time |
| Real-time capability | Yes — triggers as the user moves | No — location is fixed to the household |
| Ad format | Push notifications, display, video | Streaming video |
| Device requirement | GPS-enabled mobile device | Internet-connected TV (no GPS capability) |
| Best use case | Conquest, proximity triggers, real-time foot traffic capture | Brand awareness at home, cord-cutter reach |
| Attribution method | GPS-confirmed visit attribution | Cross-device matching, IP-based visit inference |
The most effective campaigns use both. A media buyer runs a mobile geo-fence around a competitor's location, capturing device IDs from visitors at GPS-level precision. The household graph maps those IDs to home CTV devices. The same brand message then reaches the household's connected TV — the consumer sees it at the moment of decision and again in their living room that evening. Industry practitioners call this the "surround sound" model, and CTV campaigns run alongside mobile geo-fencing consistently show stronger branded search lift than either channel in isolation.
The Real Limitations
IP Geolocation Accuracy
The 13% household linkage accuracy from the CIMM/Truthset 2025 study is the most rigorous published benchmark on the subject. Data providers agree on those linkages only 6.4% of the time. At the postal level, accuracy ranges from 5% to 17% by state. City-level remains workable at 75–85%.
"The answer changes depending on which CTV platform you're using — and the various methods and standards can be difficult to follow, even for seasoned media buyers." — IAB Tech Lab, CTV Programmatic Buyer's Guide
Dynamic IP Assignment
ISPs rotate residential IP addresses on reconnects or scheduled intervals. For CTV campaigns relying on IP-level audience lists, 30-day campaign windows with refresh cycles are the standard workaround.
VPN and CGNAT Interference
VPN usage masks the viewer's true location. Carrier-Grade NAT (CGNAT) routes multiple households through a single shared IP, inflating assumed reach counts. Platforms flag known VPN and proxy IPs in bidding logic, but no solution eliminates the problem.
IPv6 Adoption
IPv6's privacy extensions allow a single household to appear as dozens of different users within a week. IPv6 addresses are under-represented by 72% in major targeting datasets, with only 15% of commercial targeting files capturing them accurately — causing frequency capping failure, inflated reach calculations, and broken attribution.
Shared Household
A CTV device is shared by everyone in the household. A 45-year-old professional and a teenager share the same connected TV, the same IP, and the same impression count. For demographics-sensitive campaigns, layering behavioral and demographic audience data on top of geographic targeting is required to approach individual relevance.
Platform Fragmentation
A campaign across Hulu, Peacock, and Paramount+ through three separate DSPs produces three attribution reports with no shared identity layer. Frequency capping fails because the systems don't communicate — a viewer can see the same ad fifteen times with no unified cap in place. Geogrammatic stitches the attribution chain from ad impression to confirmed visit, so buyers aren't manually reconciling conflicting data sources.
How CTV Attribution Works
Step 1: Impression Logged. The platform records device ID or IP address, timestamp, creative ID, and geographic signal.
Step 2: Household Graph Link. The CTV device is matched to a household graph, connecting it to the home's mobile phones, tablets, and desktop computers.
Step 3: Cross-Device Tracking. If the household's mobile phone visits the advertiser's website, conducts a branded search, or downloads an app, the action is attributed to the CTV impression. This is how CTV captures conversions without the viewer ever clicking anything.
Step 4: Foot Traffic Attribution. Mobile phone IDs from the household graph are monitored for GPS-confirmed visits to the advertiser's store. When a device from an exposed household enters the store's geo-fence, the visit is matched back to the CTV impression.
Attribution windows vary by purchase cycle. Restaurants: 48 hours. Auto dealerships: 30 days for standard segments, extended to 60–90 days for luxury vehicles and EVs. Furniture and home goods: 90 days or longer.
What Attribution Can Measure:
- Post-view website visits
- Branded keyword search lift following ad exposure
- Mobile app downloads
- In-store foot traffic (GPS-confirmed)
- Online purchases (revenue pixel)
- CRM file matches (hashed email matched to CTV household)
Incrementality Testing
The most defensible approach withholds ads from a matched control group and measures lift among exposed versus unexposed households — separating ad-driven behavior from baseline. Geogrammatic's reporting layer connects the full attribution chain across platforms, so cross-DSP campaigns produce a single coherent answer rather than three conflicting reports.
The Platform Landscape for CTV Geo-Targeting
Platform selection has real consequences for geo-targeting precision and attribution quality.
The Trade Desk
The largest independent DSP. Deep CTV partnerships with Roku, Disney, and Netflix. Uses UID2, an encrypted email-based identity framework. Supports household-level IP targeting and first-party data onboarding with geo-targeting from DMA down to ZIP code level. Best suited for cross-channel programmatic buyers who need scale.
Amazon DSP
Walled-garden DSP with exclusive Amazon purchase and browsing data. Over 250 million Fire TV devices sold globally. Strongest for e-commerce and retail where purchase intent data adds precision beyond location alone. Amazon's closed-loop attribution connects ad exposure to purchases within its ecosystem. The limitation: cross-platform reach requires a second DSP.
Roku Ads Manager
Roku's own DSP (formerly OneView). Reaches 90 million streaming households — at least one Roku device in 50% of U.S. broadband homes. First-party ACR data feeds both targeting and measurement, which is the platform's clearest differentiation from independent DSPs. Platform revenue surpassed $1 billion in Q4 2024, up 25% year-over-year.
Xandr / Microsoft Invest
Microsoft announced the sunset of Microsoft Invest (formerly Xandr) in early 2026, citing strategic priorities in AI advertising tools. Advertisers previously using Xandr have largely migrated to The Trade Desk, DV360, or Amazon DSP.
Magnite
The largest independent SSP — supply side, not a DSP. Approximately 25% U.S. SSP market share, 99% global CTV inventory coverage. ContextIQ provides AI-powered contextual targeting at the inventory level.
The Multi-DSP Reality
Amazon and Roku push advertisers toward their own DSPs, each offering exclusive first-party data independent DSPs cannot replicate. Most sophisticated CTV buyers run a multi-DSP model: Roku Ads Manager for ACR-based audience data, The Trade Desk or Amazon DSP for cross-platform scale. The cost: fragmentation — multiple platforms, multiple reports, no unified view.
Practical Implementation: What to Do With This Information
Match targeting tier to your confidence level. DMA and city-level targeting performs reliably. ZIP code targeting is workable with demographic layering. Household-level IP targeting should be treated as directional.
Layer signals rather than relying on IP alone. Platforms that combine IP geolocation with ACR data, household graph matching, and GPS-derived addresses from mobile produce materially better household linkages.
Set campaign windows at 30 days maximum for IP-based lists. Dynamic IP rotation makes longer windows inaccurate. Refresh mappings at 30-day intervals for extended campaigns.
Run mobile geo-fencing alongside CTV, not instead of it. The surround-sound model outperforms either channel in isolation for advertisers with physical locations.
Build incrementality tests in from launch. A control group established before the campaign goes live is the only way to isolate ad-driven lift from baseline behavior.
Evaluate vendors on attribution transparency. Ask: How is foot traffic measured? What device graph do you use and what's the match rate? How do you handle cross-platform frequency capping? The answers reveal whether a platform is selling attribution or dashboard optics.
FAQ
What is CTV geo-targeting and how does it differ from traditional TV local advertising?
CTV geo-targeting delivers ads to viewers in specific geographic areas — from broad DMAs to individual ZIP codes or household addresses. Traditional linear TV locks you to DMA-level buys: you pay for the entire market regardless of audience match. CTV lets you bid on individual impressions within a DMA that meet both geographic and demographic criteria. And unlike linear, CTV connects the ad impression to downstream behavior through cross-device tracking and foot traffic measurement.
How accurate is IP address targeting on connected TV?
Less accurate than most platforms suggest. State-level is reliable. City-level drops to 75–85%. At the household level, the 2025 CIMM/Truthset study found 13% linkage accuracy — providers agree on those linkages only 6.4% of the time. Combining IP with ACR data, device graph matching, and GPS-derived household data meaningfully improves precision.
What is the difference between OTT geo-targeting and CTV geo-targeting?
OTT is the delivery method (streaming over the internet); CTV is the device category (internet-connected TV screens). OTT includes mobile, tablet, and desktop inventory — all of which can carry live GPS data. CTV targets only the TV screen, which cannot report GPS coordinates, so location is always inferred through household IP, ACR data, and device graph matching. Use mobile OTT for real-time proximity targeting. Use CTV for household-level brand reach on the big screen.
How does foot traffic attribution work for CTV campaigns?
The process: (1) a CTV ad is served and the household is logged; (2) via device graph, associated mobile phone IDs are identified; (3) when one of those phones enters the advertiser's store geo-fence (GPS-confirmed), the visit is registered; (4) the device is matched back to the exposed household and the visit is attributed to the campaign. Attribution windows vary by category: 48 hours for restaurants, 30 days for standard automotive, up to 90 days for luxury vehicles and high-consideration purchases.
Which platforms support CTV geo-targeting, and how do I choose between them?
The Trade Desk is the largest independent DSP with ZIP-code-level geo-targeting via UID2. Amazon DSP offers Fire TV access with exclusive shopper data, strongest for e-commerce. Roku Ads Manager provides first-party ACR-based household targeting across 90 million U.S. streaming households. Choose based on your primary objective: cross-platform scale (The Trade Desk), Amazon purchase data (Amazon DSP), or Roku-native accuracy (Roku Ads Manager). Most sophisticated buyers run a multi-DSP model, which makes unified attribution the operational priority.
What This Means for Your Next Campaign
CTV geo-targeting is a real, scalable channel. The opportunity — $23.6 billion in annual spend, 90% household penetration, video completion rates approaching 96% — is validated. The mechanics are more constrained than the sales pitch suggests.
13% IP-to-household accuracy does not mean CTV geo-targeting does not work. It means DMA and city-level campaigns perform with confidence, while household-level targeting requires layered signal validation. Attribution is real and measurable — foot traffic, branded search lift, post-view web visits — but only when the methodology is built in from the start.
The buyers who treat CTV as an extension of mobile geo-fencing — connecting the channels into a coherent full-funnel strategy — are the ones producing ROAS data worth presenting. The ones treating it as a replacement, or relying on IP targeting alone, are the ones with reports that raise more questions than they answer.
Geogrammatic provides geo-targeting analytics and attribution infrastructure for marketing professionals running local and regional campaigns across CTV and programmatic channels.