Deterministic data & clean rooms
Industry Discussion | Author: Anita Munro, Mindshare APAC
The changes ahead of us are for the good
With the ongoing changes to privacy and data regulation there are many types of identity solutions coming into the marketplace. It's clear that our future is likely to need us, as an advertising industry, to work across multiple solutions.
Each of these work with various types of data and make connections to that data in differing ways. First party data is becoming more and more important in this new landscape.
This industry discussion focused on the value of working specifically with deterministic first party data and clean room solutions, where we see this fitting into the ecosystem and whether it can be a scaled solution.
The changes that are happening ahead of us are fundamentally a good thing.
They provide us with a privacy first approach which allows ethical and sustainable marketing practices, and transparency and control for consumers.
Cookies and IDs were never precise, people-based or interoperable. A chance to rethink and rebuild at scale is valuable.
The industry now needs to accelerate first party approaches to data on both the advertiser and publisher side and find alternatives for activation and measurement in privacy safe ways.
Data source and type will impact media value
The biggest impact we are seeing with changes to data privacy will be on inferred or probabilistic identifiers.
We hear a lot about third party cookies and device IDs being phased out but changes are also impacting identification through IP addresses and even email addresses.
This will mean a move to a greater reliance on authenticated first party data. Consent will be a key factor to consider – how we ask for data, how we gather it, how we store it and how we act on it. The good news is we now have more time to get our alternatives and solutions right and get the learning in place.
What do these changes mean from a media value perspective?
Based on third party cookies diminishing:
We do anticipate that desktop inventory CPMs without targeting available will go down in value. Websites with highly sought after niche audiences will likely see growing demand for their inventory.
Publishers with login first party data will have a higher value. However very few publishers require registration of users at this point.
Based on Opt in for app tracking on mobile:
This will mainly impact IOS inventory where users have opted out. We will see a decrease in value there.
For now Android remains unaffected which is important in the context of APAC as a large part of app inventory is bought on the Android operating system.
But no doubt more change and more impact will come.
So what types of data will be most valuable moving forward?
There are two main types of digital identifiers, Probabilistic or Deterministic:
Probabilistic data contains passive signals with varying levels of persistence from connections, devices, apps and content that consumers access eg, cookies, device ID and IP address
Deterministic data uses IDs created and controlled by consumers, relating to accounts for services eg, email address, account ID, phone number and address.
Both types of data have value but there are some trade offs.
Deterministic data scale is still relatively small, but with high accuracy and high privacy control as opt in and consent is acquired.
Probabilistic data has more scale but is significantly less accurate and has less privacy control due to limited opt in and consent.
Data clean rooms provide a closed data sharing solution
Data clean rooms allow advertiser and publisher first party (deterministic) data to be matched and used without any exchange of the data and without democratisation of data which helps to hold data value for publishers.
The way that this works is both the publisher and the client upload their data to the clean room.
A connection is made using an ID that exists in both data sets.
Once data is connected, the advertiser can see the amount of matches and generate aggregated insights. The results can also be enriched with further data.
These matched IDs can then be activated via the ad server or DSP.
If there is no ID that matches, a bridging partner can be brought in to further connect the dots.In terms of the benefits of this approach:
It provides a closed ecosystem that enables matching of multiple data sets for mutual benefit and retained value
No data is shared - no data moves, eliminating risk of data leaks
Queries are run only to provide intersection of data owners already have
Privacy compliant as long as opt in and consent has been gained on both sides
The challenge currently is on the scalability of such solutions given that many advertisers are still building their first party data strategies, as are publishers. However as probabilistic signals continue to deteriorate we do expect that this will grow in scale and opportunity.
