With the incoming demise of the third-party cookie, identifying indivdual users is about to get a whole lot more difficult, and there are seemingly hundreds of solutions claiming to offer the panacea to fix it. Florian Lichtwald, Managing Director and Chief Business Officer at Zeotap looks at the six categories that advertisers should consider, from ‘lowest quality, highest scale’ to ‘lowest scale, highest quality’.
Addressability: put simply, the ability to identify and connect with individual users, regardless of the platform they’re on. Everyone knows that with the demise of the third-party cookie, it’s about to get a whole lot more difficult, and there are seemingly hundreds of solutions claiming to offer the panacea to fix it. From Google’s FLoC to Universal IDs, you’d be forgiven for thinking that surviving means alighting upon a silver bullet solution.
However, the new reality of addressability isn’t going to be any one of those: it’s far more likely to be a mix of all of them. Rather than success depending on hitting a bullseye on a single solution, it’s really about understanding the diversity of solutions in the market and respecting the fact that each one of them has its own strengths and weaknesses.
In the simplest form, the future of addressability involves combining solutions to create an equilibrium between scale and quality. Generally speaking, solutions show an inverse relationship between these two factors (i.e. those that can deliver scale do so with less accuracy, and vice versa). This means that advertisers need to carefully choose combinations that shore up the weaknesses of one with the strengths of the other.
The potential breadth of this landscape is intimidating from the outside, but actually fairly easily classified on further examination. Here are the six categories that advertisers should consider, from ‘lowest quality, highest scale’ to ‘lowest scale, highest quality’.
1. Contextual
The simplest to understand – along with being the most prone to error. Contextual targeting is the practice of advertising on a website that is relevant to the interests or characteristics of your audience.
For example, a brand that sells tennis equipment may want to advertise on sports websites or alongside content featuring popular players.
However, this scenario also demonstrates the limits of context, as marketers need to rely on generalizations about their audiences that reduce the effectiveness of their advertising. The scale is here, but the accuracy is limited.
2. Cohort-based and browser-based
A cohort is a large group of people with similar browsing habits, where the members of each cohort remain supposedly anonymous. This is the land of bird-based acronyms, from Google’s FLoC (Federated Learning of Cohorts) and FLEDGE (First Locally-Executed Decision over Groups Experiment), alongside offerings like SPARROW from Criteo and PARRROT from Magnite.
The principle here is important (reiterated by Google recently): the pseudonymous nature of cohorts mean that user privacy is protected and respected, while still allowing the advertiser to target. However, this creates its own challenge: you need cohorts that are sufficiently large for the purposes of privacy, but still with meaningful enough attributes for accurate personalisation. Given that this approach is browser-specific, that’s a significant hurdle.
This is a problem that is still in the process of being solved – tests for cohorts are ongoing, and many of the black box-like algorithms that drive them mean that publishers and advertisers alike might not yet be confident in what they can deliver in terms of accuracy.
3. Probabilistic
Next, we have probabilistic or fingerprinting-based advertising, which relies on the metadata drawn from a user’s device to create a profile. So if someone visits your site but is not logged in, you can use information such as that person’s browser of choice, download speeds, etc. to build a profile. This profile remains pseudonymous, but the combination of multiple signals combined with continuously evolving algorithms can be used for pretty granular-level targeting.
If done right, probabilistic identification should rely on the explicit consent of the user. This creates a problem: consent frameworks are mainly based on identifiers such as cookies or email addresses today. However, probabilistic targeting relies on background signals – this is less transparent from a user perspective, and may warrant additional communication to the user in order to be fully compliant.
So, probabilistic targeting has its limitations – not least because it relies on device information, making the cross-device journey difficult to track – but is an option that many are adopting.
4. Authenticated users
Next up is where we start to get to ‘premium quality’ – authenticated users, aka the Universal ID story. Universal IDs use persistent identifiers – hashed emails, for example – to identify users and match them to their existing user profile to ensure that marketers are transacting against the right audiences – provided, of course, that these users have consented to their data being collected.
Of the four tactics listed above, Universal IDs such as ID+ enable the most accurate option for addressability in a post-cookie world (and the most watertight with regards to consumer privacy). This is the most premium of the options we’ve listed – while it may not be able to deliver at huge scale, it ensures that one-to-one personalisation can still be achieved.
5. The walled gardens
Google, Facebook, Pinterest, Amazon, Snapchat – in the new reality of addressability, they’re not going anywhere. They have identity locked down, and they add to that with a rich pool of data about individuals – and they do all of that at scale. However, their ability to be a panacea is hampered by a lack of transparency around how data is collected, handled and shared. So while we’ll all continue to play there, we’re aware that we do so under the house’s rules – this underscores the importance of having interoperable solutions for the open web that can deliver viable alternatives.
6. First-party data
Last but not least: if the demise of third-party cookies has taught us anything, it’s how important it is to have a compliant, structured and rich pool of first-party data. This first-party data asset is how publishers can ‘build their own’ walled gardens with their high-quality authenticated data, giving advertisers a new means of high-accuracy addressability in a cookieless future. That’s why we’ve seen so many businesses onboarding customer data platforms in the last couple of years, including our own Customer Intelligence Platform, as the cookieless future has put a ticking clock on the need to unify, manage and activate this data.
This data can be transacted on through Private Marketplace (PMP) deals on a number of Supply Side Platforms (SSPs), meaning that there’s scalability in this space as well as accuracy.
In summary, the new reality of addressability is likely best approached through quantum theory, in that there is no single reality, but rather a multitude of them. This is our new compromise: while we now have to get comfortable with complexity, we do so knowing that we’re creating an ecosystem that better prioritises consumer privacy and trust.
By Florian Lichtwald
Managing Director and Chief Business Officer
Zeotap