If you’ve ever seen the same ad follow you from your phone, to your laptop, and to your smart TV, you’re not imagining things. That seamless cross-device experience is powered by identity graphs.
Identity graphs are a key technology in programmatic advertising. In this article, you'll learn what they are, how they work, and why they’re essential for modern digital marketing.
What is an identity graph?
An identity graph is a database that links all the different identifiers associated with a single person or household. For example, your identity graph entry might contain this information:
- Cookie ID from your Chrome browser
- Mobile device ID from your phone
- IP address from your home Wi-Fi
- Browsing behavior and content interactions
- AgenID (DSP-specific identifier)
- Various identity solution IDs (e.g. RampID, UID2)
- Other bid request signals (such as user-agent, location, or publisher IDs)
Think of it like a digital file folder: each piece of data on its own isn’t very useful, but when combined, the result is a coherent identity profile that enables smarter targeting, frequency management, and measurement, all while respecting privacy standards.
Why is it called a graph?
When talking about identity graphs, the word “graph” doesn’t mean a bar chart or line graph. The word is borrowed from computer science. It's a way of showing relationships between things.
- Nodes in an identity graph are identifiers, like cookies, device IDs, and email addresses.
- Edges are connections that link nodes. They show how different IDs relate to one another.
Imagine a web of connected dots, where each dot (i.e. node) is a piece of your digital identity. That’s, graphically, what an identity graph looks like behind the scenes.
How does it work?
An identity graph isn’t built out of thin air. It’s powered by a process called identity resolution. Platforms like illumin use identity resolution to match identifiers and recognize that they belong to the same user.
Identity resolution relies on two key matching techniques:
Deterministic matching
This method uses verified, one-to-one connections. For example, if you log in to the same account on your phone and laptop, the system knows both devices belong to you. This type of data is considered highly accurate and includes sources like:
- Mobile advertising ID
- CRM data
- First-party cookies
Probabilistic matching
This method uses AI and pattern analysis to make educated guesses. If two devices use the same Wi-Fi, access similar content at the same time, and behave in similar ways, the system infers they belong to the same person or household. Common sources include:
- IP address
- Device type
- Location
- Browsing behavior
Both matching techniques work together to “resolve” multiple identifiers into a single, unified profile, which is then stored in the identity graph.
Identity resolution in action
Here’s what that process looks like in a programmatic advertising environment:
- Data collection: Signals come in from websites and apps.
- Identity resolution: These signals are stitched together using deterministic and probabilistic methods to create unified user profiles.
- Activation: Demand-side platforms (DSPs) like illumin use these profiles to deliver relevant ads across devices and channels, while controlling frequency and improving measurement.
All of this happens behind the scenes and at lightning speed. Millions of identifiers are matched in milliseconds. Think of identity resolution as a marketer’s GPS. It's constantly recalculating the best route to reach the right person. The identity graph is the detailed map on the car dashboard filled with paths and signals. Together, they guide programmatic advertising with speed and accuracy.
Why are identity graphs important?
Without identity graphs, online advertisers would be driving without a GPS. Lost, off-course, and hoping for the best. They’d see your behavior on one device but would have no way to connect it to your online behavior elsewhere.
Here’s what identity graphs make possible:
- Cross-device targeting: Show someone an ad on their phone, and then retarget them later on their laptop.
- Frequency control: Avoid showing the same ad 10 times to the same person on different devices.
- Personalization: Deliver messaging based on user behavior across the entire journey.
- Measurement: Attribute a purchase on a desktop to an ad viewed on a mobile device.
In a world where users constantly jump between devices, identity graphs are essential for effective and efficient advertising.
Privacy and regulation
Identity graphs must follow strict data privacy laws. Many DSPs use hashed identifiers, and user data is pseudonymized. Consent frameworks, like GDPR in Europe and CCPA in California, limit what can be collected and how it’s used. So, while identity graphs are powerful, they also have to be privacy-first.
Key takeaways
- Identity graphs link multiple user identifiers to build a unified profile.
- They enable cross-device targeting, frequency control, and better measurement.
- They rely on both deterministic (known) and probabilistic (inferred) data.
- They must follow data privacy laws and industry best practices.
In short, identity graphs help brands find users across devices and deliver messages that feel relevant. So the next time you see an ad for those shoes you browsed on your tablet pop up on your laptop, you’ll know what’s guiding it. Welcome to the world of identity graphs, where every signal points in the right direction.
FAQs
Why do identity graphs matter for cross-device journeys?
Identity graphs matter because modern customer journeys span multiple devices and channels. Without a unified identity layer, marketers lose the ability to control frequency, sequence messages, and measure impact across the full funnel.
What happens if identity resolution accuracy is low?
Low accuracy leads to fragmented profiles and inflated reach. Marketers see higher frequency overlap, weaker attribution signals, and reduced confidence in reporting because exposures and conversions fail to connect reliably.
What are best practices for identity graph usage in a DSP?
Best practice relies on balanced deterministic and probabilistic signals, strong consent governance, and continuous model refinement. Marketers also align frequency and attribution logic to identity resolution outputs rather than device-level signals.
How does identity resolution affect measurement and attribution?
Identity resolution allows conversions on one device to link back to exposures on another device. This connection improves attribution accuracy and helps marketers evaluate true campaign impact across channels.