CTV Device Graphs Explained: Linking TVs, Phones, and Households
What a CTV device graph is, how it links TVs to other devices and households, how accuracy is tested, and how it affects measurement.
The short answer
A CTV device graph links streaming TV devices to other devices, such as phones and laptops, and to a household. It uses signals like shared IP addresses, device IDs, and login data to infer which devices belong together. Graphs power cross-device targeting, household frequency capping, and outcome measurement, but they are probabilistic in places, so their accuracy should be tested against known truth data before they drive budget decisions.
CTV measurement often needs to connect a TV ad to something that happens on another device: a website visit, a search, or an app download. The device graph is what makes that link. Its quality quietly shapes targeting, frequency, and outcome numbers.
How a device graph is built
| Signal | Link type | Strength |
|---|---|---|
| Shared login across devices | Deterministic | Strong |
| Shared IP address over time | Probabilistic | Medium; IPs change and are shared |
| Device co-location patterns | Probabilistic | Medium to weak |
| Platform account data | Deterministic within platform | Strong but limited to that platform |
Most graphs combine both types. The provider decides confidence thresholds, which changes how many devices get linked.
What the graph is used for
- Targeting. Reach a household on TV based on data tied to another device.
- Frequency. Cap across TV and other devices in the same home.
- Measurement. Link TV exposure to visits or actions on other devices.
Accuracy tradeoffs
A graph can be tuned for:
- Precision. Fewer links, but more of them correct.
- Scale. More links, more errors.
For measurement, wrong links create noise and can inflate or deflate results. For targeting, they waste impressions. Ask providers which way they tune and whether you can choose.
Testing a graph
- Truth set comparison. Compare the graph's links with a known set of verified device-household relationships.
- Holdout consistency. Check that exposed and control groups are linked at similar rates.
- Match rate by platform. Some TV platforms link much better than others.
The identity graph evaluation article gives a broader checklist.
Health considerations
Device graphs connect behavior across devices, which raises privacy concerns when health is involved. Points to check:
- Is health-related data used to build or activate graph segments?
- Do consumers have notice and choice?
- Do state health privacy laws apply?
Many health advertisers use device graphs for frequency and measurement with aggregated reporting, and avoid using them to carry health inferences across devices.
Common mistakes
- Assuming all graph links are equally reliable.
- Comparing measurement results from providers using different graphs.
- Ignoring match rate differences between TV platforms.
- Using graphs to move health inferences across devices without review.
Practical takeaway
Ask your CTV measurement provider what share of exposed households were linked to other devices, split by deterministic and probabilistic links. If most links are probabilistic, treat cross-device outcome results as directional.
Frequently asked questions
What is a device graph?
A map of which devices belong to the same person or household, built from shared signals such as IP addresses, logins, and device IDs.
How accurate are CTV device graphs?
It varies by provider and by link type. Deterministic links from logins are stronger than links inferred from shared IP addresses.
Why does a device graph matter for measurement?
Measurement often connects a TV exposure to an action on a phone or computer. If the link between devices is wrong, the measurement is wrong.
Sources
External guidance and platform documentation change. Links were current at publication; check them again before relying on them for a decision.
Editorial note. Analysis and frameworks are the author's own and do not represent Acxiom or any current or former employer, client, or named platform. Examples labeled hypothetical or illustrative are not results from real campaigns. Nothing here is legal, regulatory, or medical advice.
New pharma programmatic breakdowns, occasionally
One email when I publish something worth reading. Benchmarks, measurement teardowns, and case studies with the caveats attached. No cadence promises, no reselling your address.
Unsubscribe any time. See the privacy policy.
Working through this decision on a real plan?
I work on health and pharma data, identity, and activation, after five years running HCP and DTC programmatic agency-side. Happy to talk through how this applies to your situation.