Due to the increasing use of mobile devices and therefore further complex customer journeys, more and more advertisers are using cross-device tracking to ensure that their digital marketing efforts reaches consumers effectively.
While most marketers don’t need convincing of the importance of cross device tracking in theory, in practice there is still a lot confusion about how to use it efficiently.
Accordingly, there are a number of questions that marketers are asking about cross device tracking to ensure that its potential can be fully exploited.
WHAT TYPES OF DEVICE GRAPHS EXIST?
There are two basic approachesfor cross device tracking: deterministic vs. probabilistic. The deterministic model uses first-party data (usually login data) to establish a link between a user and his devices. If a user logs on to an advertiser's website from different devices, it is clear that the devices in question belong to the same user. Probabilistic matching, on the other hand, collects a large number of data points across each device and uses algorithms to determine likely connections. This method relies heavily on principles of probability and is therefore very inaccurate. The deterministic model, on the other hand, is very precise a connection is only detected when there is a clear link between a user and a device.
WHAT DATA IS USED FOR MATCHING?
For an exact matching of the deterministic approach, a unique identifier of the user is needed. Usually, this is existing login data, such as a customer id or an encrypted email address, which the visitor uses regularly and on different devices.
HOW ACCURATE IS THE DATA IN THE CROSS DEVICE GRAPH?
Depending on the cross device model, the accuracy of the data can vary greatly. The deterministic model relies on the user to perform a certain action on the advertiser's website, usually the login to one of his devices. This form of tracking is therefore very accurate. It is more precise than probabilistic tracking, since a connections is only made if there is a clear link between a user and a device. The probabilistic model, on the other hand, has a greater range, since it does not depend on users performing an action like a login. This method is therefore less precise than deterministic matching and the hit rates vary widely and sometimes only reach 60%.
IS THE MATCHING PRIVACY-COMPLIANT?
In order to comply with data protection guidelines, personal data that can be used to identify a person must be made anonymous. Information such as the email address must therefore always be encrypted so that the reference to a person can no longer be made. With secure encryption mechanisms, companies ensure that sensitive customer data cannot be decrypted and is therefore protected at all times.
Read more about this topic in our blog post: Data security in cross-device tracking