Consumer online activity today is fragmented across different devices and platforms. Members’ exposure to marketing on one device often results in engagement (such as a visit to the advertiser page and/or purchase of their product) on a different device.
For members who have not logged in to LinkedIn, the LinkedIn cookie is absent, so identification is not possible. For members who are not logged in to LinkedIn and are outside the Designated Countries and Brazil, we infer the association between the member and the device.
This inference is used to measure the effectiveness of ads (by attributing engagement on one device to an ad seen on another), to provide analytics that don’t identify you, and to serve relevant ads on and off LinkedIn.
Our identity graph technology does not seek to infer interests. LinkedIn Marketing Solutions only personalizes ads for our members. We do not seek to profile non-members, and we also do not create or enhance behavioral profiles of members with off-LinkedIn data.
Below are the key observation data fields collected to probabilistically infer identity.
-
Cookies on a mobile or desktop browser, Google Ad ID on Android
-
Operating system, device make and model (User Agent)
-
IP address
-
Time of access (Time Stamp)
-
Page URL or application name, as applicable
The probabilistic identity graph technology forms likely pairs/associations of IDs (such as a Google Ad ID on Android, cookies on browsers) to the same device and devices to the same user or group. More importantly, this is done such that we don’t have in this system an understanding of which specific user (as identified by their member information) or group is involved.
The observation data collected (as described above) for probabilistic identity inferences is retained for 90 days.
Data collected is immediately pseudonymized. Due to the pseudonymization, the algorithms can at best learn which devices (and identifiers) likely belong to the same person (or device) or group and not who the individual is. Pseudonymized data in the graph processing line is highly confidential and strict access control policies are enforced.
This inference of identity is used to measure the effectiveness of ads, provide analytics that do not identify you, and serve relevant ads on and off LinkedIn.
As a member, you can opt out by using the respective settings that control each of the four purposes of our inference of identity: attribution, retargeting, analytics, and off-LinkedIn ads display. In addition, both members and non-members can visit the Digital Advertising Alliance opt-out page and check the LinkedIn opt-out box.