Scott Switzer, our CTO, takes a further look at future alternative solutions in the ID space - both of which the Ozone team is actively exploring…
In our first post about ID, we discussed some of the initiatives underway at Ozone to help our advertising ecosystem navigate into a cookie-free world. In this unchartered future, identifying anonymous users and sending bid requests between ad platforms with these anonymous IDs as data keys will probably not be possible. Instead, there are two major initiatives that are most likely be used for digital ad campaigns in the future - Verified IDs and Micro-segmentation.
Ozone doesn’t expect a “winner takes all” in the post-cookie world for a number of reasons. Firstly, while certain audiences won’t mind sharing their information with our publishers, others might. Secondly, there are alternative targeting strategies - particularly for prospecting campaigns to reach new audiences - that don’t necessarily require user-level data. This means that Verified ID and Microsegmentation will work in unison.
Verified ID in focus
Verified IDs are email addresses or mobile numbers obtained by our publishers by asking their users to sign up for newsletters, to register in order to view certain content, or to subscribe and pay a monthly fee. Verified ID in some ways uses the same framework built for anonymous ID. However, there are two important differences; registration share and the limitations on trading data
1. Registration share
While publisher registrations and subscriptions are growing strongly it's unlikely the proportion of users sharing Verified IDs for advertising purposes will match those of 3rd party cookie consent. While Anonymous ID consent hovers around the 80% mark, future consent for Verified ID is as yet unknown and will be dependent on the business model of the publisher. For example, some could (and have) installed paywalls which will affect registration in a positive way.
2. Limitations on trading data
GDPR regulation applies greater controls around how personal data is shared between parties and for what purposes. Before the GDPR, email hashes without salt or encryption were passed between parties, similar to anonymous IDs, whereas this is now considered an insecure method for general targeting, and is a common area of malpractice in the ad industry.
The future approach which adheres to more of a privacy-by-design for the purpose of targeting users is to organise users into segments based on their interests. Instead of passing IDs individually between platforms, an audience segment is created with the overlapping email addresses between the advertiser and publisher, and only the segment (and not the ID) is traded between platforms. Here is a simplified data flow of verified ID between systems:
Fig 1: Simplified data flow showing Verified ID between systems
Verified ID in summary
Verified IDs allow advertisers to leverage the data that subscribers share with publishers, but without the data leakage that comes with sharing identifiers between different media companies. Additionally, Verified IDs can allow publishers to create premium audiences segments that deliver greater value to the advertiser. However, it is unlikely that the majority of a publisher’s audience will constent to Verified ID without significant changes to business models.
Micro-segmentation in focus
Ozone is also building detailed audience segments for users who choose not to give their verified ID to a publisher. This type of audience segmentation, called Micro-segmentation, groups audience members into increasingly detailed demographic, interest, and intent categories.
The IAB is moving in this direction. Their standard audience taxonomy has moved from 100+ tier 1 and 2 categories in 2015 to over 1500 categories in 6 tiers today.
Fig 2: The growing shape of a standard audience taxonomy
Facebook has taken this further to include keywords in its category hierarchy to produce thousands or millions more micro-segments.
In parallel with our own R&D into microsegmentation, Ozone is following the working groups of several other microsegmentation concepts, including Google Turtledove (building behavioural segments locally in the browser) and WFA Cross Media Measurement initiative (mapping personal behaviours to virtual users who represent segments). The key to these concepts is broad adoption by the ad tech community, so we will continue to invest time and resources into a few leading initiatives until a front-runner becomes clear.
1. Ozone can measure historical audience behaviour
Ozone uses first party IDs to measure audience behaviour across our platform. In addition, Ozone uses a variety of signals (as discussed in our earlier ID article) to measure audience across Ozone publishers. These signals are not shared with other ad platforms like anonymous IDs are today - they are used to build micro-segments. These micro-segments become the data that is shared between ad platforms in a cookie-free world.
2. Probabilistic Micro-segmentation
Ozone publishers have deterministic audience micro-segments based around interests due to the deep context that news articles provide. It is not as straightforward to determine other segments like demographics as the vast majority of publishers do not collect demographic data from their viewers. However, probabilistic micro-segmentation can be done.
Demographic segments can be derived with a high success rate using supervised machine learning models predicting demography using context and declared demographic data. Purchase intent can also be derived from Ozone audience categories in a similar fashion.
Micro-segmentation in summary
Micro-segmentation is especially interesting when derived from the data of premium content publishers like news and magazine brands. News articles, and historical viewing by Ozone’s combined audience gives incredibly strong signals toward interest categories and keywords. Context will become more and more important as cookies fade and advertisers focus on rich micro-segments as a core media channel.
