Mobile Tutorial Series: What is Cross-Device Matching / User-to-Device Matching
It’s Time to Think Mobile-First
To take a consumer- or customer- first approach to marketing, you need to understand all of a person’s digital behavior. Brands that use only PC data for targeting are missing out on understanding vast portions of the consumer’s total behavior.
In order for a brand to fully understand a consumer’s behavior, it must aggregate behavioral and transactional signals from across all of a person’s devices into one profile. That’s a task that is easier said than done.
First, PCs and mobile devices require different technologies for tracking. PC-based tracking is generally managed with cookies, whereas mobile tracking is more often achieved by associating actions and marketing activity to a device advertising ID. While it can be hard to connect cookies on multiple PCs (like a work and a home PC,) it is even more difficult to connect cookies to mobile device advertising IDs.
Two Methodologies for Linking Devices to Users
Many companies claim to link multiple devices and device types to users, but the methodologies used vary significantly, as do the veracity of the linkages. Many use the terms deterministic matching and probabilistic matching to describe their approaches.
- Deterministic Matching: Here the approaches by which users are matched to devices do so with certainty. Two devices are linked to one another because of some common input that occurs on each device. One key deterministic methodology is to connect multiple devices by a user’s common login. In this method, media companies can connect individual users to their various devices when they log on to sites from more than one screen. For example, a single login used on two PCs, a mobile phone and a tablet would provide a deterministic indication that all of those devices were used by the same person.
- Probabilistic Matching: Probabilistic matching uses a variety of non-individual-specific signals to infer connections between devices. For example, two different devices that connect on the same six WIFIs over the course of a day have a greater likelihood of being used by the same person. Home WIFI stations are particularly valuable here because fewer devices connect to them than, say, a Starbucks at 54th and Lexington Avenue in New York. The key difference between probabilistic versus deterministic is the level of certainty with which the user-to-device match is made. With probabilistic, there is always an element of uncertainty, though this can be reduced based on the methodology used to infer a connection.
Both Deterministic and Probabilistic Matching Have a Role
Best actors in probabilistic matching use other signals beyond household WIFI to define their connections. Other signals include browsing patterns, time-based clues, and device proximity. Both deterministic and probabilistic matches have their place in cross-device marketing and omni-channel profile development.
Deterministic matches are obviously preferable, but scale can be a challenge. That’s why many companies use both deterministic and probabilistic matches to get accuracy and scale.
Cross-Device Matching and Match Rates
In cross-device matching, we often talk about “match rate.” There are actually two kinds of match rate. One is the rate at which a company can match known individuals to anonymized cookie IDs. The other is the ability to match devices to one another.
As matches are made, solutions providers create a device graph of connections that can be leveraged by marketers to target the same user across many different devices.