The average American has eight different devices. Between smartphones, digital watches, laptops, tablets, gaming consoles, PCs, and smart TVs, we’re more connected than ever before.
And our data is more dispersed than ever before.
When a customer sees an advertisement on one device, does research on another, and makes a purchase with a third device, how do advertisers see the whole picture?
That’s where a device graph comes in handy.
Device graphs are data structures that anonymously link individuals to their personal devices.
Device graphs collect continuous inputs from many different data streams, and this data is categorized, organized, and validated to provide a more complete picture of each household—and the multiple devices in it.
How does a device graph work?
Imagine John and Lisa live together in a household. Combined, they own two cell phones, one Smart TV, one tablet, and one laptop. All their devices are registered to the same household, but John and Lisa don’t use each device. John uses his phone, the tablet, and the TV, and Lisa uses her phone, her laptop, and the TV.
To decipher which devices belong to John and Lisa individually, a device graph aggregates feeds from many different data sources (browser activity, mobile GPS pings, and more), then identifies John and Lisa as unique users and links each device to its true user(s).
The data contained in a device graph that’s used for device linking comes from many different sources. A few examples include:
Data, by itself, isn’t that useful. It’s a lot of numerical values that don’t really mean anything unless you can identify what it is and why it’s important. Device graph identifiers provide the organization and context that make this data useful.
Identifiers are nodes of data that have a relationship. Common device graph identifiers include:
Individual nodes of data on different identifiers have limited usefulness. Knowing what devices were connected during an ad play—but not where those devices were located—is like putting together a puzzle that is missing pieces.
A device graph fills in the gaps by using identity links to connect identifiers and show relationships.
It’s easy to misinterpret data taken out of context. To ensure accuracy and reliability, each input is validated or double checked in a handful of ways.
For example, a validation sequence might check how often a specific device ID is associated with a certain geolocation tag or IP address. By doing this, outliers or guest devices are not mistaken as a part of the household.
Another example might be a validation to see if there is a pattern between device IDs and the networks they attach to throughout the day. This data can help identify differences between device use cases. A work device would likely be connected to a company broadband network or a public wifi access point during business hours, while a personal device would connect to a home wifi network in the evening.
Once the data is collected, organized, linked, and has passed several thousand rounds of validation–it becomes usable. At this point, advertisers can pull the data they need by performing searches using an interfacing console.
How this works varies from one provider to another. Currently, Madhive is the only OTT-first device graph that links audiences to the devices they use in truly real-time.
Device graphs are especially useful for targeting the right audiences with relevant messaging. They’re also integral to post-campaign attribution.
Let’s say you’re a sports fan who’s always researching the latest news and refreshing game scores on your phone. But when you’re sitting on the couch watching CTV content, you see a fashion ad.
Sounds like a mismatch, right?
With the help of a device graph, the fashion advertiser would know you’re the user for both your mobile device and your TV, so chances are you’re not an ideal customer. In your place, the advertiser would target a CTV viewer who has a clearer interest in fashion across their devices.
Imagine you’re a retailer who wants to know your CTV campaign’s effectiveness at driving in-store traffic. The device graph makes it possible.
First, you’d create a geofence using the coordinates of your storefront. Then, when your campaign is live, you’d record the mobile device IDs of everyone who crosses that geofence.
Next, you’d use a device graph to cross-reference the mobile IDs with other devices belonging to the same users. If your device graph finds you served someone an ad on one of their devices and they subsequently visited your storefront, you’d know the ad was a success.
When it comes to full-funnel attribution, the industry has more work to do to turn TV advertising into a true performance channel. Device graphs are the first step. When you factor in the device data marketers now have access to, along with smart TV data — aka ACR data — and other measurement tools like panels, TV has the ability to become a true closed-loop performance vehicle.
Device graphs might get a bad rep for being complex data analytics tools. But here’s what you really need to know — they’re just well-designed data warehouses designed to make audience information more useful.
Device graphs make it possible to target precise audiences with relevant ads, then accurately measure campaigns across devices.
You might also like:
